JapanKanji

1
{
2
 "cells": [
3
  {
4
   "cell_type": "code",
5
   "execution_count": 1,
6
   "metadata": {},
7
   "outputs": [],
8
   "source": [
9
    "import datetime\n",
10
    "import sys, os\n",
11
    "\n",
12
    "import cv2 as cv2\n",
13
    "import numpy as np\n",
14
    "import pandas as pd\n",
15
    "import matplotlib.pyplot as plt\n",
16
    "import skimage.transform"
17
   ]
18
  },
19
  {
20
   "cell_type": "code",
21
   "execution_count": 2,
22
   "metadata": {},
23
   "outputs": [],
24
   "source": [
25
    "sys.path.append('../Python')"
26
   ]
27
  },
28
  {
29
   "cell_type": "code",
30
   "execution_count": 3,
31
   "metadata": {},
32
   "outputs": [],
33
   "source": [
34
    "from xyCreate import xTrainCreate, yTrainCreate3, drawTrain1, drawPred1\n",
35
    "from formSubs import s, softmax, formSub, formSub1, formSub1x4, drawSub1A, drawSub1F"
36
   ]
37
  },
38
  {
39
   "cell_type": "code",
40
   "execution_count": 4,
41
   "metadata": {},
42
   "outputs": [],
43
   "source": [
44
    "dirData   = '../Data'\n",
45
    "dataTrain = os.path.join(dirData,'train_images')\n",
46
    "dataTest  = os.path.join(dirData,'test_images')\n",
47
    "dirTrain  = dataTrain\n",
48
    "dirTest   = dataTest\n",
49
    "\n",
50
    "dataSave  = '../Work'\n",
51
    "dirSave   = dataSave\n",
52
    "dirLoad   = dirSave\n",
53
    "\n",
54
    "dirPython = '../Python'\n",
55
    "dirModels = '../Model'\n",
56
    "dirResul  = '../Result'\n",
57
    "dirLogs   = '../Logs'"
58
   ]
59
  },
60
  {
61
   "cell_type": "code",
62
   "execution_count": 5,
63
   "metadata": {},
64
   "outputs": [],
65
   "source": [
66
    "#sys.path.append(dirPython)"
67
   ]
68
  },
69
  {
70
   "cell_type": "code",
71
   "execution_count": null,
72
   "metadata": {},
73
   "outputs": [],
74
   "source": []
75
  },
76
  {
77
   "cell_type": "code",
78
   "execution_count": null,
79
   "metadata": {},
80
   "outputs": [],
81
   "source": []
82
  },
83
  {
84
   "cell_type": "code",
85
   "execution_count": null,
86
   "metadata": {},
87
   "outputs": [],
88
   "source": []
89
  },
90
  {
91
   "cell_type": "code",
92
   "execution_count": 6,
93
   "metadata": {},
94
   "outputs": [
95
    {
96
     "data": {
97
      "text/html": [
98
       "<div>\n",
99
       "<style scoped>\n",
100
       "    .dataframe tbody tr th:only-of-type {\n",
101
       "        vertical-align: middle;\n",
102
       "    }\n",
103
       "\n",
104
       "    .dataframe tbody tr th {\n",
105
       "        vertical-align: top;\n",
106
       "    }\n",
107
       "\n",
108
       "    .dataframe thead th {\n",
109
       "        text-align: right;\n",
110
       "    }\n",
111
       "</style>\n",
112
       "<table border=\"1\" class=\"dataframe\">\n",
113
       "  <thead>\n",
114
       "    <tr style=\"text-align: right;\">\n",
115
       "      <th></th>\n",
116
       "      <th>image_id</th>\n",
117
       "      <th>labels</th>\n",
118
       "    </tr>\n",
119
       "  </thead>\n",
120
       "  <tbody>\n",
121
       "    <tr>\n",
122
       "      <th>0</th>\n",
123
       "      <td>100241706_00004_2</td>\n",
124
       "      <td>U+306F 1231 3465 133 53 U+304C 275 1652 84 69 ...</td>\n",
125
       "    </tr>\n",
126
       "    <tr>\n",
127
       "      <th>1</th>\n",
128
       "      <td>100241706_00005_1</td>\n",
129
       "      <td>U+306F 1087 2018 103 65 U+304B 1456 1832 40 73...</td>\n",
130
       "    </tr>\n",
131
       "    <tr>\n",
132
       "      <th>2</th>\n",
133
       "      <td>100241706_00005_2</td>\n",
134
       "      <td>U+306F 572 1376 125 57 U+306E 1551 2080 69 68 ...</td>\n",
135
       "    </tr>\n",
136
       "    <tr>\n",
137
       "      <th>3</th>\n",
138
       "      <td>100241706_00006_1</td>\n",
139
       "      <td>U+3082 1455 3009 65 44 U+516B 1654 1528 141 75...</td>\n",
140
       "    </tr>\n",
141
       "    <tr>\n",
142
       "      <th>4</th>\n",
143
       "      <td>100241706_00007_2</td>\n",
144
       "      <td>U+309D 1201 2949 27 33 U+309D 1196 1539 27 36 ...</td>\n",
145
       "    </tr>\n",
146
       "    <tr>\n",
147
       "      <th>5</th>\n",
148
       "      <td>100241706_00008_1</td>\n",
149
       "      <td>U+25B2 1056 929 96 39 U+309D 379 1098 21 43 U+...</td>\n",
150
       "    </tr>\n",
151
       "    <tr>\n",
152
       "      <th>6</th>\n",
153
       "      <td>100241706_00008_2</td>\n",
154
       "      <td>U+25B2 1648 955 95 44 U+25B2 1887 947 96 45 U+...</td>\n",
155
       "    </tr>\n",
156
       "    <tr>\n",
157
       "      <th>7</th>\n",
158
       "      <td>100241706_00009_1</td>\n",
159
       "      <td>U+3078 1551 2071 104 41 U+3078 323 1473 135 43...</td>\n",
160
       "    </tr>\n",
161
       "    <tr>\n",
162
       "      <th>8</th>\n",
163
       "      <td>100241706_00009_2</td>\n",
164
       "      <td>U+309D 1452 1423 20 37 U+3078 690 2535 121 41 ...</td>\n",
165
       "    </tr>\n",
166
       "    <tr>\n",
167
       "      <th>9</th>\n",
168
       "      <td>100241706_00010_1</td>\n",
169
       "      <td>U+3078 587 1127 100 40 U+3064 1103 1480 60 45 ...</td>\n",
170
       "    </tr>\n",
171
       "  </tbody>\n",
172
       "</table>\n",
173
       "</div>"
174
      ],
175
      "text/plain": [
176
       "            image_id                                             labels\n",
177
       "0  100241706_00004_2  U+306F 1231 3465 133 53 U+304C 275 1652 84 69 ...\n",
178
       "1  100241706_00005_1  U+306F 1087 2018 103 65 U+304B 1456 1832 40 73...\n",
179
       "2  100241706_00005_2  U+306F 572 1376 125 57 U+306E 1551 2080 69 68 ...\n",
180
       "3  100241706_00006_1  U+3082 1455 3009 65 44 U+516B 1654 1528 141 75...\n",
181
       "4  100241706_00007_2  U+309D 1201 2949 27 33 U+309D 1196 1539 27 36 ...\n",
182
       "5  100241706_00008_1  U+25B2 1056 929 96 39 U+309D 379 1098 21 43 U+...\n",
183
       "6  100241706_00008_2  U+25B2 1648 955 95 44 U+25B2 1887 947 96 45 U+...\n",
184
       "7  100241706_00009_1  U+3078 1551 2071 104 41 U+3078 323 1473 135 43...\n",
185
       "8  100241706_00009_2  U+309D 1452 1423 20 37 U+3078 690 2535 121 41 ...\n",
186
       "9  100241706_00010_1  U+3078 587 1127 100 40 U+3064 1103 1480 60 45 ..."
187
      ]
188
     },
189
     "execution_count": 6,
190
     "metadata": {},
191
     "output_type": "execute_result"
192
    }
193
   ],
194
   "source": [
195
    "trainY = pd.read_csv('../Data/train.csv')\n",
196
    "trainY.head(10)"
197
   ]
198
  },
199
  {
200
   "cell_type": "code",
201
   "execution_count": 7,
202
   "metadata": {},
203
   "outputs": [
204
    {
205
     "data": {
206
      "text/html": [
207
       "<div>\n",
208
       "<style scoped>\n",
209
       "    .dataframe tbody tr th:only-of-type {\n",
210
       "        vertical-align: middle;\n",
211
       "    }\n",
212
       "\n",
213
       "    .dataframe tbody tr th {\n",
214
       "        vertical-align: top;\n",
215
       "    }\n",
216
       "\n",
217
       "    .dataframe thead th {\n",
218
       "        text-align: right;\n",
219
       "    }\n",
220
       "</style>\n",
221
       "<table border=\"1\" class=\"dataframe\">\n",
222
       "  <thead>\n",
223
       "    <tr style=\"text-align: right;\">\n",
224
       "      <th></th>\n",
225
       "      <th>symb</th>\n",
226
       "      <th>x</th>\n",
227
       "      <th>y</th>\n",
228
       "      <th>w</th>\n",
229
       "      <th>h</th>\n",
230
       "    </tr>\n",
231
       "  </thead>\n",
232
       "  <tbody>\n",
233
       "    <tr>\n",
234
       "      <th>0</th>\n",
235
       "      <td>U+306F</td>\n",
236
       "      <td>1231</td>\n",
237
       "      <td>3465</td>\n",
238
       "      <td>133</td>\n",
239
       "      <td>53</td>\n",
240
       "    </tr>\n",
241
       "    <tr>\n",
242
       "      <th>1</th>\n",
243
       "      <td>U+304C</td>\n",
244
       "      <td>275</td>\n",
245
       "      <td>1652</td>\n",
246
       "      <td>84</td>\n",
247
       "      <td>69</td>\n",
248
       "    </tr>\n",
249
       "    <tr>\n",
250
       "      <th>2</th>\n",
251
       "      <td>U+3044</td>\n",
252
       "      <td>1495</td>\n",
253
       "      <td>1218</td>\n",
254
       "      <td>143</td>\n",
255
       "      <td>69</td>\n",
256
       "    </tr>\n",
257
       "    <tr>\n",
258
       "      <th>3</th>\n",
259
       "      <td>U+3051</td>\n",
260
       "      <td>220</td>\n",
261
       "      <td>3331</td>\n",
262
       "      <td>53</td>\n",
263
       "      <td>91</td>\n",
264
       "    </tr>\n",
265
       "    <tr>\n",
266
       "      <th>4</th>\n",
267
       "      <td>U+306B</td>\n",
268
       "      <td>911</td>\n",
269
       "      <td>1452</td>\n",
270
       "      <td>61</td>\n",
271
       "      <td>92</td>\n",
272
       "    </tr>\n",
273
       "  </tbody>\n",
274
       "</table>\n",
275
       "</div>"
276
      ],
277
      "text/plain": [
278
       "     symb     x     y    w   h\n",
279
       "0  U+306F  1231  3465  133  53\n",
280
       "1  U+304C   275  1652   84  69\n",
281
       "2  U+3044  1495  1218  143  69\n",
282
       "3  U+3051   220  3331   53  91\n",
283
       "4  U+306B   911  1452   61  92"
284
      ]
285
     },
286
     "execution_count": 7,
287
     "metadata": {},
288
     "output_type": "execute_result"
289
    }
290
   ],
291
   "source": [
292
    "#\n",
293
    "#  heights & weights\n",
294
    "#\n",
295
    "aa=[np.array(l.split()).reshape((-1,5)) for l in trainY.dropna().labels]\n",
296
    "bb=np.vstack(aa)\n",
297
    "\n",
298
    "dfClasses         = pd.DataFrame()\n",
299
    "dfClasses['symb'] = bb[:,0]\n",
300
    "dfClasses['x']    = bb[:,1].astype(np.int16)\n",
301
    "dfClasses['y']    = bb[:,2].astype(np.int16)\n",
302
    "dfClasses['w']    = bb[:,3].astype(np.int16)\n",
303
    "dfClasses['h']    = bb[:,4].astype(np.int16)\n",
304
    "\n",
305
    "dfClasses.head()"
306
   ]
307
  },
308
  {
309
   "cell_type": "code",
310
   "execution_count": 8,
311
   "metadata": {},
312
   "outputs": [
313
    {
314
     "data": {
315
      "text/html": [
316
       "<div>\n",
317
       "<style scoped>\n",
318
       "    .dataframe tbody tr th:only-of-type {\n",
319
       "        vertical-align: middle;\n",
320
       "    }\n",
321
       "\n",
322
       "    .dataframe tbody tr th {\n",
323
       "        vertical-align: top;\n",
324
       "    }\n",
325
       "\n",
326
       "    .dataframe thead th {\n",
327
       "        text-align: right;\n",
328
       "    }\n",
329
       "</style>\n",
330
       "<table border=\"1\" class=\"dataframe\">\n",
331
       "  <thead>\n",
332
       "    <tr style=\"text-align: right;\">\n",
333
       "      <th></th>\n",
334
       "      <th>symb</th>\n",
335
       "      <th>hcount</th>\n",
336
       "      <th>hmean</th>\n",
337
       "      <th>hstd</th>\n",
338
       "      <th>wcount</th>\n",
339
       "      <th>wmean</th>\n",
340
       "      <th>wstd</th>\n",
341
       "    </tr>\n",
342
       "  </thead>\n",
343
       "  <tbody>\n",
344
       "    <tr>\n",
345
       "      <th>80</th>\n",
346
       "      <td>U+306B</td>\n",
347
       "      <td>24685</td>\n",
348
       "      <td>81.266964</td>\n",
349
       "      <td>23.367150</td>\n",
350
       "      <td>24685</td>\n",
351
       "      <td>66.787766</td>\n",
352
       "      <td>24.607299</td>\n",
353
       "    </tr>\n",
354
       "    <tr>\n",
355
       "      <th>83</th>\n",
356
       "      <td>U+306E</td>\n",
357
       "      <td>24136</td>\n",
358
       "      <td>79.463913</td>\n",
359
       "      <td>23.876515</td>\n",
360
       "      <td>24136</td>\n",
361
       "      <td>82.174428</td>\n",
362
       "      <td>23.512733</td>\n",
363
       "    </tr>\n",
364
       "    <tr>\n",
365
       "      <th>60</th>\n",
366
       "      <td>U+3057</td>\n",
367
       "      <td>22209</td>\n",
368
       "      <td>124.199514</td>\n",
369
       "      <td>50.427857</td>\n",
370
       "      <td>22209</td>\n",
371
       "      <td>33.656761</td>\n",
372
       "      <td>27.332906</td>\n",
373
       "    </tr>\n",
374
       "    <tr>\n",
375
       "      <th>75</th>\n",
376
       "      <td>U+3066</td>\n",
377
       "      <td>20569</td>\n",
378
       "      <td>95.881278</td>\n",
379
       "      <td>24.440175</td>\n",
380
       "      <td>20569</td>\n",
381
       "      <td>67.325247</td>\n",
382
       "      <td>21.920820</td>\n",
383
       "    </tr>\n",
384
       "    <tr>\n",
385
       "      <th>77</th>\n",
386
       "      <td>U+3068</td>\n",
387
       "      <td>16588</td>\n",
388
       "      <td>83.424463</td>\n",
389
       "      <td>21.888026</td>\n",
390
       "      <td>16588</td>\n",
391
       "      <td>43.707499</td>\n",
392
       "      <td>13.900656</td>\n",
393
       "    </tr>\n",
394
       "  </tbody>\n",
395
       "</table>\n",
396
       "</div>"
397
      ],
398
      "text/plain": [
399
       "      symb  hcount       hmean       hstd  wcount      wmean       wstd\n",
400
       "80  U+306B   24685   81.266964  23.367150   24685  66.787766  24.607299\n",
401
       "83  U+306E   24136   79.463913  23.876515   24136  82.174428  23.512733\n",
402
       "60  U+3057   22209  124.199514  50.427857   22209  33.656761  27.332906\n",
403
       "75  U+3066   20569   95.881278  24.440175   20569  67.325247  21.920820\n",
404
       "77  U+3068   16588   83.424463  21.888026   16588  43.707499  13.900656"
405
      ]
406
     },
407
     "execution_count": 8,
408
     "metadata": {},
409
     "output_type": "execute_result"
410
    }
411
   ],
412
   "source": [
413
    "dfCount = dfClasses[['symb','h','w']].groupby(['symb']).agg(['count','mean','std'])\n",
414
    "dfCount = dfCount.reset_index()\n",
415
    "dfCount.columns = ['symb','hcount','hmean','hstd','wcount','wmean','wstd']\n",
416
    "dfCount = dfCount.fillna(0)\n",
417
    "dfCount.sort_values(['hcount'], ascending=False).head()"
418
   ]
419
  },
420
  {
421
   "cell_type": "code",
422
   "execution_count": 9,
423
   "metadata": {},
424
   "outputs": [
425
    {
426
     "data": {
427
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA30AAAEWCAYAAADSJv4JAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvqOYd8AAAIABJREFUeJzt3XucZVV95/3P125RvCDgpR/STdIYO0aEEbUHUTRWJEIrSTATnQeGCWBI+omBRA2TBDOTMN6ewdxQEzVhhADGiIoaOogiA1TUKFdRLiKhxY50QJFwkTZe0uY3f+xV9OniVNWpS9fl1Of9ep1X7b322vus/etTtfq399rrpKqQJEmSJA2nRyx0AyRJkiRJu45JnyRJkiQNMZM+SZIkSRpiJn2SJEmSNMRM+iRJkiRpiJn0SZIkSdIQM+mTJEmSpCFm0qclLckbklw8ruy2CcqO7rP/i5JsG/eqJL/Ytv/FuG3fT/LgBG35iSQXJvlWknuTXJLk6ePqvD7JN5I8kOTsJI/qc5wXtza8pafsgHa8e5JM+eWaSY5Ocmt7n7uTnJtkj57teyf5WJLvJPmnJP9lgGPuluQrSbaOK1+R5C1J7kzyYJLrk+w5YDtOTnJti+s5U7VBkjR9SbYk+ZlxZSck+ewk+7wgydXt7/oNSV44Qb2/an3W0yY51s8luan1o59Lsn/Ptqn6ib9OcleSbyf5xyS/Mu7Yv5Jkczv2J5P8yCTt2L/1Ofe11//pbUtPvb793STHfVgMkjwjyeXtvDYn+YVB253kt1u8HkzytSS/PUg7pMmY9Gmp+zRwaJIVAEn+H+CRwHPGlT2t1d1JVX2mqh439gJ+FtgGfLJt/7Vx2z8AfHiCtuwJbAKeDqwCrgYuHNuY5AjgVOAwYC3wVOCNvQdI8kjgHcBV4479b8CHgBMHiAnAPwCHVtUT2vusBN7Ss/1dwA9aO48F3pPkmVMc87eBu/uUvxF4AfB8YA/gl4DvDdiOO9v62QOelyRpF0uyN11/9kd0fdsfAn+XZK9x9V4I/PgUx1oHvB/4tXasvwM2JVnZqkzVT/wvYG1V7QH8PPCWJM9tx34x8P8DRwF7A1+j66cncifwylb3Se0cz+9Tb6L+rt/5PSwG7dwuBC5q77UR+OskPzFguwMcB+wFbABOTp8L19J0mPRpqbuGLsk7qK3/FHAFcOu4sq9W1Z0DHO944IKq+s74DUkeC/wicG6/Havq6qo6q6rurap/A84Anp7kiT3HPquqbq6q+4A3AyeMO8wpwKeAr4w79q1VdRZw8wDnQFXdUVX39BT9kC7x7T2P36+qbVX1WbqO75cmOl6S/YD/Stf59pbvBbwO+NWq+qfq3FRV35uqHW37R6vqb4F/GeS8JEnz4gXAN6vqw1X1w6r6a+BbwH8aq9ASmz8DTp7iWEcAn6mqz1bVduBtwGrgxTBQP3FzVX1/bLW9xpKsnwM+3Or8gK5f/akkfRPRqrq/qrZUVdElVju9Vzuvvv1dP5PE4CeBHwHOaPG7nC65HetnJ213Vf1hVX2hqrZX1a10CeShU7VHmoxJn5a09sfyKrrEjvbzM8Bnx5U97C7feEkeQ3cFsG9SR5cofWuQY/W87zeqaiyheSbwpZ7tXwJWjSWFSX4M+GXgTQMef1JJXpjkAeDB1va3t00/Afywqv5xXFsmu9P3Z8DvAd8dV34gsB14Zbphq/+Y5KQB2yFJWpzSXuPLDuhZfz3w6aq6YZrHGlt/6FhT9RNJ3p3kX+kuiN4FXDzuWL3HZlw7H96g5H66ESl/RnfHrddE/V0/E8VgfOzGyg7oWR6o3UkCvIgBL/pKEzHp0zD4e3YkeC+iS/o+M67s7wc4zi8C90xS93jgvHaFcFJJ1tANofytnuLHAQ/0rI8tP779fCft7tsAbZ1Su6r6BGAN3RCdLRO0Y6wtj6eP9hzCyqr6WJ/Na4An0CWS+9Elzf8zyUsHaIckaf78bZL7x17Auyep+zngR5Ick+SRSY6nu7v2GIAk+wL/H/AHA7zvpcCLk4wk2Y0uodpt7FgwdT9RVb9O10e9CPgoMHbn72LgPyf5D0l2b+2p3mP3U1V70vVdJwPXj5VP0d/tZIoYfIVueOhvt/gdTndnc6xd02n3/6T7//pfTdUmaTImfRoGnwZe2IYaPrmqbqPrsF7Qyg5gsLtzEyZ17Y/7i4HzpjpIkifTDdF8d1X1jtHfRvfM25ix5QeT/Bzw+Kr64ADtHP9+x2bHRDOfGL+9qv6Z7hnFsecWxrdjrC0Pm6CmDQX9Q+A3Jnj7sSuhb6qq77arnecDLx+gHZKk+fOKqtpz7AX8+kQV2wiVo+guXH6T7rmy/wOMTWzydrq/++MvIPY71lfo+tc/p7tL9yTgyz3H6q07YT/Rhkl+li4xfE0ruww4DfgI8E90yeKDwNYkP9rTNz7sYmp7jOMvgPOSPGWA/m68CWPQHvF4BXAk8A26Rzc+NHbOk7W79zhJTqZ7tu/IniGu0oyY9GkYfJ7uit1GujHzVNW36R7Y3gjcWVVfm+wALakbYeKk7jjgc1V1+xTH2Ysu4dtUVW8dt/lm4Fk968+ie2biX+gmd1nfhkh+A/h/gdcluZApVNX7eyabedkE1Vay4xmIfwRWtofre9vSb+jIOrpJZz7T2vVRYJ/WzrXA2JCWKe9+9mmHJGmRqqq/r6r/WFV70z2L9nS6Ccqg67P+qKfPAvh8JpgJuqouqKoDquqJdMnOj9E9k9/PVP3ETtur6l1Vta6qnkKXRK0Ebqqqr4+biK2fR9DdXVvN1P3deJPGoKpuqKoXV9UTq+oIuklqxuI3YbvHtif5Zdrkb1U10Cyi0mRM+rTkVdV3gWvprkh+pmfTZ1vZIHf5fokuqfvqBNuPA86Z7ADpppi+BPiHqjq1T5XzgBPTTRm9F/A/eo75+3RDJA9qr03A/wZe3Y6dJI+mGxJDkkenz9c99LTl2HaVM+1ZwbcCl8FDVzc/CrwpyWOTHEp3Rfd9fQ51E7BvT7t+he6q70HAHS1enwH+e5JHJXkGXcJ60VTtaNtXtvNaAaxo57USSdKCSvLsNjRxD+CPga1VdUnb/BN0FwvH+gboJifpOywyyXPTfb3Pk4G/BP6u3QGctJ9od+COTvK4tv8RwDHA5W37o9N9pVGS/ChwJvCONllav3a8tJ3XinZefwrcB9zCFP1dn8NNGoM2dPPRSR6T5L8B+9D6/KnaneRYumcNXzrVxWZpUCZ9GhZ/DzyFLtEb85lWNkjSdxwTTOCS5Pl0w0ke9lUNST6R5Pfa6i8A/xF4dXb+br8fBaiqT9INHbmCbjjHP9Fd8aSqHqyqb4y96IZNfqeq7m3H/rFWNnY37rt0M5ROZH+6Ia7b6O5+3gr8as/2Xwd2p3vm4APAa6rq5nZOLxobCtNmDutt173Av7f1H7ZjHdPa9y/Ax+meSxxL7KZqx/9o53Iq3Wxp321lkqSF9Tt0z7nfQZewPPQ9c1V197i+AeCedhF2fN8I3VcR3U/XB9zPzv3AZP1E0Q3l3EqXnP0x8LqqGhsF82jgb9q+V9ON/Pn9Sc5pT7o+7wHgq3Qzd26oqu8N0t+1Pv1Fg8SA7mLyXXT97GF0CdzYEM2p2v0W4InANT3/l/iLSc5LmlIGmJNCkiRJkrREeadPkiRJkoaYSZ8kSZIkDTGTPkmSJEkaYiZ9kiRJkjTEluzU6E960pNq7dq109rnO9/5Do997GN3TYOWGGOxg7HYmfHYwVjssJCxuO666+6pqicvyJsvQTPpH3v5ue/PuPRnXPozLv0Zl/5mE5dB+8glm/StXbuWa6+9dlr7jI6OMjIysmsatMQYix2Mxc6Mxw7GYoeFjEWSf1qQN16iZtI/9vJz359x6c+49Gdc+jMu/c0mLoP2kQ7vlCRJkqQhZtInSZIkSUPMpE+SJEmShphJnyRJkiQNMZM+SZIkSRpiJn2SJEmSNMRM+iRJkiRpiJn0SZIkSdIQM+mTJEmSpCG2cqEboJlbe+rHd1rfcvqRC9QSSZKmp7cPs/+SpF3LO32SJEmSNMRM+iRJkiRpiJn0SZIkSdIQM+mTJEmSpCFm0idJkiRJQ8ykT5IkSZKGmEmfJEmSJA0xkz5JkiRJGmImfZIkSZI0xAZK+pJsSXJjki8mubaV7Z3k0iS3tZ97tfIkeWeSzUluSPKcnuMc3+rfluT4nvLntuNvbvtmrk9UkiRJkpaj6dzp++mqOqiq1rf1U4HLqmodcFlbB3gZsK69NgLvgS5JBE4DngccDJw2lii2Oht79tsw4zOSJEmSJD1kNsM7jwLObcvnAq/oKT+vOlcCeybZBzgCuLSq7q2q+4BLgQ1t2x5V9fmqKuC8nmNJkiRJkmZh0KSvgE8luS7Jxla2qqruAmg/n9LKVwN39Oy7tZVNVr61T7kkSZIkaZZWDljv0Kq6M8lTgEuTfGWSuv2ex6sZlD/8wF3CuRFg1apVjI6OTtro8bZt2zbtfRazUw7cvtP6dM5t2GIxG8ZiZ8ZjB2Oxg7GQJGnpGijpq6o728+7k3yM7pm8bybZp6ruakM0727VtwL79uy+BrizlY+MKx9t5Wv61O/XjjOBMwHWr19fIyMj/apNaHR0lOnus5idcOrHd1rfcuzIwPsOWyxmw1jszHjsYCx2MBaSJC1dUw7vTPLYJI8fWwYOB24CNgFjM3AeD1zYljcBx7VZPA8BHmjDPy8BDk+yV5vA5XDgkrbtwSSHtFk7j+s5liRJkiRpFga507cK+Fj7FoWVwN9U1SeTXAN8KMmJwNeBV7X6FwMvBzYD/wq8GqCq7k3yZuCaVu9NVXVvW34NcA6wO/CJ9pIkScvA2vEjV04/coFaIknDacqkr6puB57Vp/xfgMP6lBdw0gTHOhs4u0/5tcABA7RXkiRJkjQNg07kIkmSNC967/x510+SZm8239MnSZIkSVrkTPokSZIkaYiZ9EmSJEnSEDPpkyRJkqQhZtInSZIkSUPMpE+SJEmShphJnyRJM5Tk9UluTnJTkg8keXSS/ZJcleS2JB9Mslur+6i2vrltX9tznDe08luTHNFTvqGVbU5y6vyfoSRpGJj0SZI0A0lWA78JrK+qA4AVwNHA24AzqmodcB9wYtvlROC+qnoacEarR5L9237PBDYA706yIskK4F3Ay4D9gWNaXUmSpsWkT5KkmVsJ7J5kJfAY4C7gJcAFbfu5wCva8lFtnbb9sCRp5edX1fer6mvAZuDg9tpcVbdX1Q+A81tdSZKmZeVCN0BzZ+2pH39oecvpRy5gSyRp+FXVPyf5Y+DrwHeBTwHXAfdX1fZWbSuwui2vBu5o+25P8gDwxFZ+Zc+he/e5Y1z58/q1JclGYCPAqlWrGB0dnfF5bdu2bVb7D+qUA7dPXQnmpS2DmK+4LDXGpT/j0p9x6W8+4mLSJ0nSDCTZi+7O237A/cCH6YZijldju0ywbaLyfqNxqk8ZVXUmcCbA+vXra2RkZLKmT2p0dJTZ7D+oE3ouVE5my7Eju7YhA5qvuCw1xqU/49KfcelvPuLi8E5JkmbmZ4CvVdW3qurfgI8CLwD2bMM9AdYAd7blrcC+AG37E4B7e8vH7TNRuSRJ02LSJ0nSzHwdOCTJY9qzeYcBXwauAF7Z6hwPXNiWN7V12vbLq6pa+dFtds/9gHXA1cA1wLo2G+hudJO9bJqH85IkDRmHd0qSNANVdVWSC4AvANuB6+mGWH4cOD/JW1rZWW2Xs4D3JdlMd4fv6Hacm5N8iC5h3A6cVFU/BEhyMnAJ3cygZ1fVzfN1fpKk4WHSJ0nSDFXVacBp44pvp5t5c3zd7wGvmuA4bwXe2qf8YuDi2bd04a0d8Bk+SdLcM+lbYuw0JUmSJE2HSd+QGp8c+hUOkiRJ0vLkRC6SJEmSNMRM+iRJkiRpiJn0SZIkSdIQM+mTJEmSpCFm0idJkiRJQ8ykT5IkSZKGmEmfJEmSJA0xkz5JkiRJGmImfZIkSZI0xEz6JEmSJGmIrVzoBkiSJE1k7akff2h5y+lHLmBLJGnpMulb5Ho7O0mSJEmaroGHdyZZkeT6JBe19f2SXJXktiQfTLJbK39UW9/ctq/tOcYbWvmtSY7oKd/QyjYnOXXuTk+SJEmSlrfpPNP3WuCWnvW3AWdU1TrgPuDEVn4icF9VPQ04o9Ujyf7A0cAzgQ3Au1siuQJ4F/AyYH/gmFZXkiRJkjRLAyV9SdYARwLvbesBXgJc0KqcC7yiLR/V1mnbD2v1jwLOr6rvV9XXgM3Awe21uapur6ofAOe3upIkSZKkWRr0mb63A78DPL6tPxG4v6q2t/WtwOq2vBq4A6Cqtid5oNVfDVzZc8zefe4YV/68fo1IshHYCLBq1SpGR0cHbH5n27Zt095noZ1y4PapKw1g/HkvxVjsKsZiZ8ZjB2Oxg7GQJGnpmjLpS/KzwN1VdV2SkbHiPlVrim0Tlfe721h9yqiqM4EzAdavX18jIyP9qk1odHSU6e6z0E6Yo4lcthw7stP6UozFrmIsdmY8djAWOxgLSZKWrkHu9B0K/HySlwOPBvagu/O3Z5KV7W7fGuDOVn8rsC+wNclK4AnAvT3lY3r3mahckiRJkjQLUz7TV1VvqKo1VbWWbiKWy6vqWOAK4JWt2vHAhW15U1unbb+8qqqVH91m99wPWAdcDVwDrGuzge7W3mPTnJydJEmSJC1zs/mevt8Fzk/yFuB64KxWfhbwviSb6e7wHQ1QVTcn+RDwZWA7cFJV/RAgycnAJcAK4OyqunkW7ZIkSZIkNdNK+qpqFBhty7fTzbw5vs73gFdNsP9bgbf2Kb8YuHg6bZEkSZIkTW0639MnSZIkSVpiTPokSZIkaYjN5pk+7SJr5+hrGiY65pbTj5zz40uSJElanLzTJ0mSJElDzKRPkiRJkoaYSZ8kSZIkDTGTPkmSJEkaYiZ9kiRJkjTETPokSZIkaYj5lQ270PivXvCrEiRJkiTNN+/0SZIkSdIQM+mTJEmSpCFm0idJ0gwl2TPJBUm+kuSWJM9PsneSS5Pc1n7u1eomyTuTbE5yQ5Ln9Bzn+Fb/tiTH95Q/N8mNbZ93JslCnKckaWkz6ZMkaebeAXyyqn4SeBZwC3AqcFlVrQMua+sALwPWtddG4D0ASfYGTgOeBxwMnDaWKLY6G3v22zAP5yRJGjImfZIkzUCSPYCfAs4CqKofVNX9wFHAua3aucAr2vJRwHnVuRLYM8k+wBHApVV1b1XdB1wKbGjb9qiqz1dVAef1HEuSpIE5e6ckSTPzVOBbwF8leRZwHfBaYFVV3QVQVXcleUqrvxq4o2f/ra1ssvKtfcofJslGujuCrFq1itHR0Rmf1LZt22a1/0ROOXD7rI+xK9o1qF0Vl6XOuPRnXPozLv3NR1xM+haB8V/tMB/vd8qB2zmhva9fJSFJM7ISeA7wG1V1VZJ3sGMoZz/9nserGZQ/vLDqTOBMgPXr19fIyMgkzZjc6Ogos9l/IifMQV+35diR2TdkhnZVXJY649KfcenPuPQ3H3Ex6Vsg853oSZLm3FZga1Vd1dYvoEv6vplkn3aXbx/g7p76+/bsvwa4s5WPjCsfbeVr+tSXJGlafKZPkqQZqKpvAHckeXorOgz4MrAJGJuB83jgwra8CTiuzeJ5CPBAGwZ6CXB4kr3aBC6HA5e0bQ8mOaTN2nlcz7EkSRqYd/okSZq53wDen2Q34Hbg1XQXVD+U5ETg68CrWt2LgZcDm4F/bXWpqnuTvBm4ptV7U1Xd25ZfA5wD7A58or0kSZoWk7555JBOSRouVfVFYH2fTYf1qVvASRMc52zg7D7l1wIHzLKZkqRlzuGdkiRJkjTETPokSZIkaYiZ9EmSJEnSEDPpkyRJkqQhZtInSZIkSUPM2TslSdIu4azVkrQ4eKdPkiRJkoaYSZ8kSZIkDTGTPkmSJEkaYlMmfUkeneTqJF9KcnOSN7by/ZJcleS2JB9Mslsrf1Rb39y2r+051hta+a1Jjugp39DKNic5de5PU5IkSZKWp0Hu9H0feElVPQs4CNiQ5BDgbcAZVbUOuA84sdU/Ebivqp4GnNHqkWR/4GjgmcAG4N1JViRZAbwLeBmwP3BMqytJkiRJmqUpk77qbGurj2yvAl4CXNDKzwVe0ZaPauu07YclSSs/v6q+X1VfAzYDB7fX5qq6vap+AJzf6kqSJEmSZmmgr2xod+OuA55Gd1fuq8D9VbW9VdkKrG7Lq4E7AKpqe5IHgCe28it7Dtu7zx3jyp83QTs2AhsBVq1axejo6CDNf8i2bdumvc9snHLg9qkrLZBVu+9o33zGZDGa78/FYmc8djAWOxgLSZKWroGSvqr6IXBQkj2BjwHP6Fet/cwE2yYq73e3sfqUUVVnAmcCrF+/vkZGRiZv+Dijo6NMd5/ZOGERfz/RKQdu509u7P75txw7srCNWWDz/blY7IzHDsZiB2OhxWD89/5tOf3IBWqJJC0t05q9s6ruB0aBQ4A9k4wljWuAO9vyVmBfgLb9CcC9veXj9pmoXJIkSZI0S4PM3vnkdoePJLsDPwPcAlwBvLJVOx64sC1vauu07ZdXVbXyo9vsnvsB64CrgWuAdW020N3oJnvZNBcnJ0mSJEnL3SDDO/cBzm3P9T0C+FBVXZTky8D5Sd4CXA+c1eqfBbwvyWa6O3xHA1TVzUk+BHwZ2A6c1IaNkuRk4BJgBXB2Vd08Z2coSZIkScvYlElfVd0APLtP+e10M2+OL/8e8KoJjvVW4K19yi8GLh6gvZIkSZKkaZjWM32SJEmSpKVloNk7Ndx6Z0NzJjRJkiRpuHinT5IkSZKGmEmfJEmSJA0xkz5JkiRJGmImfZIkSZI0xEz6JEmSJGmIOXvnHOudCVOSJEmSFpp3+iRJkiRpiJn0SZIkSdIQM+mTJEmSpCFm0idJkiRJQ8ykT5IkSZKGmEmfJEmSJA0xkz5JkiRJGmImfZIkSZI0xEz6JEmSJGmImfRJkiRJ0hAz6ZMkaRaSrEhyfZKL2vp+Sa5KcluSDybZrZU/qq1vbtvX9hzjDa381iRH9JRvaGWbk5w63+cmSRoOJn2SJM3Oa4FbetbfBpxRVeuA+4ATW/mJwH1V9TTgjFaPJPsDRwPPBDYA726J5ArgXcDLgP2BY1pdSZKmxaRPkqQZSrIGOBJ4b1sP8BLgglblXOAVbfmotk7bflirfxRwflV9v6q+BmwGDm6vzVV1e1X9ADi/1ZUkaVpWLnQDtLisPfXjO61vOf3IBWqJJC0Jbwd+B3h8W38icH9VbW/rW4HVbXk1cAdAVW1P8kCrvxq4sueYvfvcMa78ef0akWQjsBFg1apVjI6OzviEtm3bNqv9e51y4PapK83CXLVzEHMZl2FiXPozLv0Zl/7mIy4mfZIkzUCSnwXurqrrkoyMFfepWlNsm6i832ic6lNGVZ0JnAmwfv36GhkZ6VdtIKOjo8xm/14njLuQONe2HDuyS4/fay7jMkyMS3/GpT/j0t98xMWkT5KkmTkU+PkkLwceDexBd+dvzyQr292+NcCdrf5WYF9ga5KVwBOAe3vKx/TuM1G5JEkD85k+SZJmoKreUFVrqmot3UQsl1fVscAVwCtbteOBC9vyprZO2355VVUrP7rN7rkfsA64GrgGWNdmA92tvcemeTg1SdKQ8U6fJElz63eB85O8BbgeOKuVnwW8L8lmujt8RwNU1c1JPgR8GdgOnFRVPwRIcjJwCbACOLuqbp7XM5EkDQWTPkmSZqmqRoHRtnw73cyb4+t8D3jVBPu/FXhrn/KLgYvnsKmSpGXI4Z2SJEmSNMRM+iRJkiRpiE2Z9CXZN8kVSW5JcnOS17byvZNcmuS29nOvVp4k70yyOckNSZ7Tc6zjW/3bkhzfU/7cJDe2fd7ZvqxWkiRpQmtP/fhDL0nSxAa507cdOKWqngEcApyUZH/gVOCyqloHXNbWAV5GN/PYOrovin0PdEkicBrdF8seDJw2lii2Oht79tsw+1OTJEmSJE05kUtV3QXc1ZYfTHILsBo4Chhp1c6le4D9d1v5eW0a6iuT7Jlkn1b30qq6FyDJpcCGJKPAHlX1+VZ+HvAK4BNzc4q7llcXJUmSJC1m05q9M8la4NnAVcCqlhBSVXcleUqrthq4o2e3ra1ssvKtfcr7vf9GujuCrFq1itHR0ek0n23btk17n6mccuD2OT3efFm1+2Btn+t4LUa74nOxlBmPHYzFDsZCkqSla+CkL8njgI8Ar6uqb0/y2F2/DTWD8ocXVp0JnAmwfv36GhkZmaLVOxsdHWW6+0zlhCV6p++UA7fzJzcO8M9/43ceWtxy+pG7sEULZ1d8LpYy47GDsdjBWEiStHQNNHtnkkfSJXzvr6qPtuJvtmGbtJ93t/KtwL49u68B7pyifE2fckmSJEnSLA0ye2eAs4BbqupPezZtAsZm4DweuLCn/Lg2i+chwANtGOglwOFJ9moTuBwOXNK2PZjkkPZex/UcS5IkSZI0C4MM7zwU+CXgxiRfbGW/B5wOfCjJicDXgVe1bRcDLwc2A/8KvBqgqu5N8mbgmlbvTWOTugCvAc4BdqebwGVJTOIiSZIkSYvdILN3fpb+z90BHNanfgEnTXCss4Gz+5RfCxwwVVskSZIkSdMzrdk7JUmSJuLXGEnS4jTQRC6SJEmSpKXJpE+SJEmShphJnyRJkiQNMZM+SZIkSRpiJn2SJEmSNMRM+iRJkiRpiJn0SZIkSdIQM+mTJEmSpCFm0idJkiRJQ8ykT5IkSZKGmEmfJEmSJA0xkz5JkiRJGmImfZIkSZI0xEz6JEmSJGmImfRJkiRJ0hBbudANWIrWnvrxhW6CJEmSJA3EO32SJEmSNMRM+iRJkiRpiJn0SZIkSdIQ85k+SZK05I1/3n7L6UcuUEskafHxTp8kSZIkDTGTPkmSJEkaYiZ9kiRJkjTETPokSZqBJPsmuSLJLUluTvLaVr53kkuT3NZ+7tXKk+SdSTYnuSHJc3qOdXyrf1uS43vKn5vkxrbPO5Nk/s9UkrTUmfRpYGtP/fhDL0kS24FTquoZwCHASUn2B04FLquqdcBlbR3gZcC69toIvAe6JBE4DXgecDBw2lii2OqDJ842AAAPZ0lEQVRs7NlvwzyclyRpyJj0SZI0A1V1V1V9oS0/CNwCrAaOAs5t1c4FXtGWjwLOq86VwJ5J9gGOAC6tqnur6j7gUmBD27ZHVX2+qgo4r+dYkiQNzK9skCRplpKsBZ4NXAWsqqq7oEsMkzylVVsN3NGz29ZWNln51j7l/d5/I90dQVatWsXo6OiMz2Xbtm0z3v+UA7fP+H3n2mxi0M9s4jLMjEt/xqU/49LffMTFpE+SpFlI8jjgI8Drqurbkzx2129DzaD84YVVZwJnAqxfv75GRkamaPXERkdHmen+Jyyi4f9bjh2Z0+PNJi7DzLj0Z1z6My79zUdcHN4pSdIMJXkkXcL3/qr6aCv+ZhuaSft5dyvfCuzbs/sa4M4pytf0KZckaVqmTPqSnJ3k7iQ39ZQ5M5kkaVlr/dVZwC1V9ac9mzYBY/3c8cCFPeXHtb7yEOCBNgz0EuDwJHu1/vRw4JK27cEkh7T3Oq7nWJIkDWyQO33n8PDZwpyZTJK03B0K/BLwkiRfbK+XA6cDL01yG/DStg5wMXA7sBn438CvA1TVvcCbgWva602tDOA1wHvbPl8FPjEfJyZJGi5TPtNXVZ9uD6j3OgoYacvnAqPA79IzMxlwZZKxmclGaDOTASQZm5lslDYzWSsfm5nMTk2StKhV1Wfp/9wdwGF96hdw0gTHOhs4u0/5tcABs2imJEkznshl3mcmg9nPTjbTmXFu/OcHdlo/5cBpH2LRWbX77GZZG6aZl5xJamfGYwdjsYOxkCRp6Zrr2Tt32cxkMPvZyWY6M85imo1srpxy4Hb+5MaZ//PP9axoC8mZpHZmPHYwFjsYC0mSlq6Zzt7pzGSSJEmStATMNOlzZjJJkiRJWgKmHN+X5AN0E7E8KclWulk4Twc+lORE4OvAq1r1i4GX080y9q/Aq6GbmSzJ2Mxk8PCZyc4BdqebwMVJXCRJkiRpjgwye+cxE2xyZrJlbO245xy3nH7kArVEkrSQxvcHi0Vvu+yjJC13Mx3eKUmSJElaAuZ69s6hsVivXEqSJEnSdHinT5IkSZKGmEmfJEmSJA0xkz5JkiRJGmImfZIkSZI0xEz6JEmSJGmImfRJkiRJ0hDzKxs0J/wSXEmSJGlx8k6fJEmSJA0xkz5JkiRJGmImfZIkSZI0xHymT5IkDbXe587BZ88lLT/e6ZMkSZKkIeadPs05r6hKkiRJi4d3+iRJkiRpiHmnrxl/d0qSJEmShoF3+iRJkiRpiHmnT7tc711Un++TJEmS5pdJnyRJWla8GClpuXF4pyRJkiQNMe/0aV75dQ6SJEnS/DLp04JyiI0kLS3DNtu1FyMlLQcO75QkSZKkIWbSJ0mSJElDzOGdkiRJjY8dSBpGJn1aNHyuQpIkSZp7Jn1atLzaKkmLw7BN3jKo3vM+5cDtjCxcUyRpVkz6JEmSBuDFSElLlUmfloTJrjLb8UqSJEkTWzRJX5INwDuAFcB7q+r0Xf2ey3W4yrAxIZQ0zBaif9TU7HskLSWLIulLsgJ4F/BSYCtwTZJNVfXlhW2ZljqH4khayhaqf/Si6OyYEEpabBZF0gccDGyuqtsBkpwPHAWY9GnOTNQJn3Lgdk6Yo//g2JlLmmP2j0NmLhJq+xpJ07VYkr7VwB0961uB542vlGQjsLGtbkty6zTf50nAPTNq4ZD5TWPxkLmMRd42F0dZcH42djAWOyxkLH5sgd53MZiv/rGXn/s+FlO/ucj6mkUTl0XGuPRnXPqbTVwG6iMXS9KXPmX1sIKqM4EzZ/wmybVVtX6m+w8TY7GDsdiZ8djBWOxgLBbMvPSPO72h/9Z9GZf+jEt/xqU/49LffMTlEbvy4NOwFdi3Z30NcOcCtUWSpMXC/lGSNGuLJem7BliXZL8kuwFHA5sWuE2SJC00+0dJ0qwtiuGdVbU9ycnAJXRTUp9dVTfvgreak6EvQ8JY7GAsdmY8djAWOxiLBTCP/WMv/637My79GZf+jEt/xqW/XR6XVD3s0QBJkiRJ0pBYLMM7JUmSJEm7gEmfJEmSJA2xZZH0JdmQ5NYkm5OcutDtmQ9Jzk5yd5Kbesr2TnJpktvaz71aeZK8s8XnhiTPWbiWz70k+ya5IsktSW5O8tpWvuzikeTRSa5O8qUWize28v2SXNVi8cE2YQRJHtXWN7ftaxey/btCkhVJrk9yUVtfzrHYkuTGJF9Mcm0rW3a/J8vVcuwrx9hPTM6/kw+XZM8kFyT5SvvcPN/PCyR5ffsduinJB9r/O5bd5yVz9P/wJMe3+rclOX42bRr6pC/JCuBdwMuA/YFjkuy/sK2aF+cAG8aVnQpcVlXrgMvaOnSxWddeG4H3zFMb58t24JSqegZwCHBS+wwsx3h8H3hJVT0LOAjYkOQQ4G3AGS0W9wEntvonAvdV1dOAM1q9YfNa4Jae9eUcC4CfrqqDer4vaDn+niw7y7ivHGM/MTn/Tj7cO4BPVtVPAs+ii8+y/rwkWQ38JrC+qg6gm3zqaJbn5+UcZvn/8CR7A6cBzwMOBk4bSxRnpKqG+gU8H7ikZ/0NwBsWul3zdO5rgZt61m8F9mnL+wC3tuW/BI7pV28YX8CFwEuXezyAxwBfaH9M7gFWtvKHfmfoZgx8flte2eplods+hzFY0/7wvgS4iO6LsJdlLNp5bQGeNK5sWf+eLJfXcu4rJ4iH/cSOc/Pv5MNjsgfwtfHnttw/L8Bq4A5g7/bvfxFwxHL9vDDL/4cDxwB/2VO+U73pvob+Th87PoBjtray5WhVVd0F0H4+pZUvmxi1oQPPBq5imcajDdP5InA3cCnwVeD+qtreqvSe70OxaNsfAJ44vy3epd4O/A7w7239iSzfWAAU8Kkk1yXZ2MqW5e/JMuS/Z2M/8TD+nXy4pwLfAv6qDXt9b5LHssw/L1X1z8AfA18H7qL7978OPy9jpvv5mNPPzXJI+tKnzO+p2NmyiFGSxwEfAV5XVd+erGqfsqGJR1X9sKoOort6ezDwjH7V2s+hjUWSnwXurqrreov7VB36WPQ4tKqeQzfU5KQkPzVJ3eUQj+XEf0/sJ8bz7+SEVgLPAd5TVc8GvsOOoXr9LIu4tKGHRwH7AT8CPJauPxlvuX1epjJRHOY0Pssh6dsK7Nuzvga4c4HastC+mWQfgPbz7lY+9DFK8ki6jvz9VfXRVrxs4wFQVfcDo3TPr+yZZGXb1Hu+D8WibX8CcO/8tnSXORT4+SRbgPPphi69neUZCwCq6s72827gY3QXBZb178kysuz/Pe0n+vLvZH9bga1VdVVbv4AuCVzun5efAb5WVd+qqn8DPgq8AD8vY6b7+ZjTz81ySPquAda1mYN2o3ugdNMCt2mhbALGZv45nu6ZhbHy49rsQYcAD4zdfh4GSQKcBdxSVX/as2nZxSPJk5Ps2ZZ3p/sDfQtwBfDKVm18LMZi9Erg8moDy5e6qnpDVa2pqrV0fxcur6pjWYaxAEjy2CSPH1sGDgduYhn+nixTy7qvtJ/oz7+T/VXVN4A7kjy9FR0GfJll/nmhG9Z5SJLHtN+psbgs689Lj+l+Pi4BDk+yV7uLengrm5mFfshxPl7Ay4F/pHt26b8vdHvm6Zw/QDee+t/orhScSDdO+jLgtvZz71Y3dLO2fRW4kW7WpQU/hzmMxQvpboffAHyxvV6+HOMB/Afg+haLm4A/aOVPBa4GNgMfBh7Vyh/d1je37U9d6HPYRXEZAS5azrFo5/2l9rp57G/lcvw9Wa6v5dhX9py7/cTUMVr2fyfHxeMg4Nr2mflbYC8/LwXwRuAr7f8Y7wMetRw/L8zR/8OBX27x2Qy8ejZtSjugJEmSJGkILYfhnZIkSZK0bJn0SZIkSdIQM+mTJEmSpCFm0idJkiRJQ8ykT5IkSZKGmEmftAslOSPJ63rWL0ny3p71P0nyW5Ps/7kB3mNLkif1KR9J8oKZtFuSpPmQZNu49ROS/PkU+/x8klOnqDOS5KIJtr0uyWOm31pp6TLpk3atzwEvAEjyCOBJwDN7tr8A+IeJdq6q2SRtI2PvLUnSsKiqTVV1+iwO8TrApE/LikmftGv9AzsSr2fSfVnpg0n2SvIo4BnA9Ul+O8k1SW5I8saxnceugCZ5RJJ3J7k5yUVJLk7yyp73+Y0kX0hyY5KfTLIW+DXg9Um+mORFSV6V5KYkX0ry6Xk4d0mSZizJk5N8pPWP1yQ5tJU/dDcwyY8nubJtf9O4O4ePS3JBkq8keX86vwn8CHBFkiuSrEhyTusfb0zy+gU4VWmXW7nQDZCGWVXdmWR7kh+lS/4+D6wGng88ANxAd0duHXAwEGBTkp+qqt7E7D8Ba4EDgacAtwBn92y/p6qek+TXgf9WVb+S5C+AbVX1xwBJbgSOqKp/TrLnLjtpSZIGt3uSL/as7w1sasvvAM6oqs+2fvQSuoulvd4BvKOqPpDk18ZtezbdBdc76S7CHlpV72yPVfx0Vd2T5LnA6qo6AMD+UcPKO33Srjd2t28s6ft8z/rngMPb63rgC8BP0iWBvV4IfLiq/r2qvgFcMW77R9vP6+iSw4nacU6SXwVWzOJ8JEmaK9+tqoPGXsAf9Gz7GeDPW1K4CdgjyePH7f984MNt+W/Gbbu6qrZW1b8DX6R//3g78NQkf5ZkA/DtWZ6PtCh5p0/a9cae6zuQbnjnHcApdB3L2XR3+v5XVf3lJMfIFO/x/fbzh0zwe11Vv5bkecCRwBeTHFRV/zLoSUiSNM8eATy/qr7bW5hM1SU+5Ps9y337x6q6L8mzgCOAk4D/DPzyjForLWLe6ZN2vX8Afha4t6p+WFX3AnvSXZ38PN1wlV9O8jiAJKuTPGXcMT4L/GJ7tm8VXaI4lQeBh66IJvnxqrqqqv4AuAfYd5bnJUnSrvQp4OSxlSQH9alzJfCLbfnoAY/7UP/YZr9+RFV9BPh94Dkzbq20iJn0SbvejXSzdl45ruyBqrqnqj5FNyTl8+25uwvoSdaajwBb6e4U/iVwFd0zgZP5O+AXxiZyAf6oPaR+E/Bp4EuzPC9Jknal3wTWt0nOvkw3Qdl4rwN+K8nVwD5M3TcCnAl8IskVdM/Zj7YhpOcAb5iTlkuLTKpqodsgaQBJHldV25I8Ebia7oH0byx0uyRJWijt+/a+W1WV5GjgmKo6aqHbJS02PtMnLR0XtVnFdgPebMInSRLPpZvsJcD9+Dye1Jd3+iRJkiRpiPlMnyRJkiQNMZM+SZIkSRpiJn2SJEmSNMRM+iRJkiRpiJn0SZIkSdIQ+7+BCXiodrRLxQAAAABJRU5ErkJggg==\n",
428
      "text/plain": [
429
       "<Figure size 1080x288 with 2 Axes>"
430
      ]
431
     },
432
     "metadata": {
433
      "needs_background": "light"
434
     },
435
     "output_type": "display_data"
436
    }
437
   ],
438
   "source": [
439
    "plt.figure(figsize=(15,4))\n",
440
    "plt.subplot(121);plt.hist(dfClasses.w, bins=100)\n",
441
    "plt.title('W {:}-{:}'.format(round(dfClasses.w.mean(),4),round(dfClasses.w.std(),4)))\n",
442
    "plt.xlabel('Weights')\n",
443
    "plt.grid()\n",
444
    "plt.subplot(122);plt.hist(dfClasses.h, bins=100)\n",
445
    "plt.title('H {:}-{:}'.format(round(dfClasses.h.mean(),4),round(dfClasses.h.std(),4)))\n",
446
    "plt.xlabel('Heights')\n",
447
    "plt.grid()\n",
448
    "plt.show()"
449
   ]
450
  },
451
  {
452
   "cell_type": "code",
453
   "execution_count": null,
454
   "metadata": {},
455
   "outputs": [],
456
   "source": []
457
  },
458
  {
459
   "cell_type": "code",
460
   "execution_count": 10,
461
   "metadata": {},
462
   "outputs": [],
463
   "source": [
464
    "if 0 :\n",
465
    "    yx, xe, ye = np.histogram2d(dfClasses.w,dfClasses.h,bins=100)\n",
466
    "    plt.figure(figsize=(15,15))\n",
467
    "    plt.imshow(yx.T,extent=[xe[0],xe[-1],ye[-1],ye[0]])\n",
468
    "    plt.yticks(np.arange(ye[-1],ye[0],-50))\n",
469
    "    plt.xlabel('Weight')\n",
470
    "    plt.ylabel('Height')\n",
471
    "    plt.grid()\n",
472
    "    plt.colorbar()\n",
473
    "    plt.show()"
474
   ]
475
  },
476
  {
477
   "cell_type": "code",
478
   "execution_count": 11,
479
   "metadata": {},
480
   "outputs": [
481
    {
482
     "data": {
483
      "text/plain": [
484
       "(6007.0078125, 683464)"
485
      ]
486
     },
487
     "execution_count": 11,
488
     "metadata": {},
489
     "output_type": "execute_result"
490
    }
491
   ],
492
   "source": [
493
    "len(dfClasses)*96*96/(1024*1024),len(dfClasses)"
494
   ]
495
  },
496
  {
497
   "cell_type": "code",
498
   "execution_count": 12,
499
   "metadata": {},
500
   "outputs": [
501
    {
502
     "name": "stdout",
503
     "output_type": "stream",
504
     "text": [
505
      "(4212,)\n"
506
     ]
507
    },
508
    {
509
     "data": {
510
      "text/plain": [
511
       "array(['U+306B', 'U+306E', 'U+3057', 'U+3066', 'U+3068', 'U+3092',\n",
512
       "       'U+306F', 'U+304B', 'U+308A', 'U+306A', 'U+3082', 'U+3044',\n",
513
       "       'U+308B', 'U+3089', 'U+305F', 'U+304F', 'U+3078', 'U+307E',\n",
514
       "       'U+304D', 'U+308C'], dtype=object)"
515
      ]
516
     },
517
     "execution_count": 12,
518
     "metadata": {},
519
     "output_type": "execute_result"
520
    }
521
   ],
522
   "source": [
523
    "classes_all = dfCount.sort_values('hcount',ascending=False)[['symb']].values.squeeze()\n",
524
    "print(classes_all.shape)\n",
525
    "classes_all[:20]"
526
   ]
527
  },
528
  {
529
   "cell_type": "code",
530
   "execution_count": 13,
531
   "metadata": {},
532
   "outputs": [
533
    {
534
     "data": {
535
      "text/plain": [
536
       "4212"
537
      ]
538
     },
539
     "execution_count": 13,
540
     "metadata": {},
541
     "output_type": "execute_result"
542
    }
543
   ],
544
   "source": [
545
    "classes = classes_all[0:2000]\n",
546
    "classes = classes_all\n",
547
    "#classes = []\n",
548
    "len(classes)"
549
   ]
550
  },
551
  {
552
   "cell_type": "code",
553
   "execution_count": null,
554
   "metadata": {},
555
   "outputs": [],
556
   "source": []
557
  },
558
  {
559
   "cell_type": "code",
560
   "execution_count": 14,
561
   "metadata": {},
562
   "outputs": [],
563
   "source": [
564
    "########################################################"
565
   ]
566
  },
567
  {
568
   "cell_type": "code",
569
   "execution_count": 15,
570
   "metadata": {},
571
   "outputs": [],
572
   "source": [
573
    "shape_net = np.array((1024,     512+256, 1))\n",
574
    "shape_net = np.array((1024+512, 512+512, 1))\n",
575
    "shape_net = np.array((1024, 1024+256, 1))\n",
576
    "\n",
577
    "n_anchors = 2\n",
578
    "\n",
579
    "dirSaveX  = os.path.join(dirSave,'Split','{shape[0]}x{shape[1]}x{shape[2]}'.format(shape=shape_net)); \n",
580
    "dirLoadX  = dirSaveX"
581
   ]
582
  },
583
  {
584
   "cell_type": "code",
585
   "execution_count": 16,
586
   "metadata": {},
587
   "outputs": [
588
    {
589
     "data": {
590
      "text/plain": [
591
       "'../Work/Split/1024x1280x1'"
592
      ]
593
     },
594
     "execution_count": 16,
595
     "metadata": {},
596
     "output_type": "execute_result"
597
    }
598
   ],
599
   "source": [
600
    "dirSaveX"
601
   ]
602
  },
603
  {
604
   "cell_type": "code",
605
   "execution_count": 17,
606
   "metadata": {},
607
   "outputs": [
608
    {
609
     "name": "stdout",
610
     "output_type": "stream",
611
     "text": [
612
      "2019-09-13 04:07:19.820684 ../Work/Split/1024x1280x1\n",
613
      "2019-09-13 04:07:30.571298\n"
614
     ]
615
    }
616
   ],
617
   "source": [
618
    "if 1 :\n",
619
    "    try : del xtrain\n",
620
    "    except : abc = 1\n",
621
    "\n",
622
    "    print(datetime.datetime.now(),dirSaveX)\n",
623
    "    trainFiles     = np.load(os.path.join(dirSaveX,'xtrain_files.npy'))\n",
624
    "    scale_all      = np.load(os.path.join(dirSaveX,'xtrain_scale.npy'))\n",
625
    "    shape_in_all   = np.load(os.path.join(dirSaveX,'xtrain_shape_in.npy'))\n",
626
    "    \n",
627
    "    split          = np.load(os.path.join(dirSaveX,'split.npy'))\n",
628
    "    anchors_net    = np.load(os.path.join(dirSaveX,'anchors_net.npy'))\n",
629
    "\n",
630
    "    xdeltas        = np.load(os.path.join(dirSaveX,'xtrain_xdeltas.npy'))\n",
631
    "    \n",
632
    "    xtrain         = np.load(os.path.join(dirSaveX,'xtrain.npy'))\n",
633
    "    print(datetime.datetime.now())"
634
   ]
635
  },
636
  {
637
   "cell_type": "code",
638
   "execution_count": 18,
639
   "metadata": {},
640
   "outputs": [],
641
   "source": [
642
    "split          = np.load(os.path.join(dirSave,'split.npy'))"
643
   ]
644
  },
645
  {
646
   "cell_type": "code",
647
   "execution_count": 19,
648
   "metadata": {},
649
   "outputs": [
650
    {
651
     "name": "stdout",
652
     "output_type": "stream",
653
     "text": [
654
      "[3383  487  297 2388  304  401   31 2576 1576 1583 2139 1597 2317 3407\n",
655
      "  845 1940 3237 2100 1607  942]\n"
656
     ]
657
    }
658
   ],
659
   "source": [
660
    "try :\n",
661
    "    trainYY = trainY.iloc[split]\n",
662
    "    print(split[:20])\n",
663
    "except :\n",
664
    "    print('split is undefied')\n",
665
    "    trainYY = trainY\n",
666
    "    split   = np.arange(len(trainY)).astype(np.int32)"
667
   ]
668
  },
669
  {
670
   "cell_type": "code",
671
   "execution_count": 20,
672
   "metadata": {},
673
   "outputs": [
674
    {
675
     "data": {
676
      "text/html": [
677
       "<div>\n",
678
       "<style scoped>\n",
679
       "    .dataframe tbody tr th:only-of-type {\n",
680
       "        vertical-align: middle;\n",
681
       "    }\n",
682
       "\n",
683
       "    .dataframe tbody tr th {\n",
684
       "        vertical-align: top;\n",
685
       "    }\n",
686
       "\n",
687
       "    .dataframe thead th {\n",
688
       "        text-align: right;\n",
689
       "    }\n",
690
       "</style>\n",
691
       "<table border=\"1\" class=\"dataframe\">\n",
692
       "  <thead>\n",
693
       "    <tr style=\"text-align: right;\">\n",
694
       "      <th></th>\n",
695
       "      <th>image_id</th>\n",
696
       "      <th>labels</th>\n",
697
       "    </tr>\n",
698
       "  </thead>\n",
699
       "  <tbody>\n",
700
       "    <tr>\n",
701
       "      <th>3383</th>\n",
702
       "      <td>hnsd012-039</td>\n",
703
       "      <td>NaN</td>\n",
704
       "    </tr>\n",
705
       "    <tr>\n",
706
       "      <th>487</th>\n",
707
       "      <td>100249537_00065_1</td>\n",
708
       "      <td>U+309D 1603 2599 25 32 U+306F 1054 551 96 47 U...</td>\n",
709
       "    </tr>\n",
710
       "    <tr>\n",
711
       "      <th>297</th>\n",
712
       "      <td>100249416_00027_1</td>\n",
713
       "      <td>U+4E00 1535 395 99 12 U+4E00 1578 967 65 13 U+...</td>\n",
714
       "    </tr>\n",
715
       "    <tr>\n",
716
       "      <th>2388</th>\n",
717
       "      <td>200021802-00048_1</td>\n",
718
       "      <td>U+4E00 912 2256 131 34 U+306F 920 2648 87 45 U...</td>\n",
719
       "    </tr>\n",
720
       "    <tr>\n",
721
       "      <th>304</th>\n",
722
       "      <td>100249416_00030_2</td>\n",
723
       "      <td>U+4E00 1717 1024 57 15 U+4E00 1665 2711 97 19 ...</td>\n",
724
       "    </tr>\n",
725
       "  </tbody>\n",
726
       "</table>\n",
727
       "</div>"
728
      ],
729
      "text/plain": [
730
       "               image_id                                             labels\n",
731
       "3383        hnsd012-039                                                NaN\n",
732
       "487   100249537_00065_1  U+309D 1603 2599 25 32 U+306F 1054 551 96 47 U...\n",
733
       "297   100249416_00027_1  U+4E00 1535 395 99 12 U+4E00 1578 967 65 13 U+...\n",
734
       "2388  200021802-00048_1  U+4E00 912 2256 131 34 U+306F 920 2648 87 45 U...\n",
735
       "304   100249416_00030_2  U+4E00 1717 1024 57 15 U+4E00 1665 2711 97 19 ..."
736
      ]
737
     },
738
     "execution_count": 20,
739
     "metadata": {},
740
     "output_type": "execute_result"
741
    }
742
   ],
743
   "source": [
744
    "trainYY.head()"
745
   ]
746
  },
747
  {
748
   "cell_type": "code",
749
   "execution_count": null,
750
   "metadata": {},
751
   "outputs": [],
752
   "source": []
753
  },
754
  {
755
   "cell_type": "code",
756
   "execution_count": null,
757
   "metadata": {},
758
   "outputs": [],
759
   "source": []
760
  },
761
  {
762
   "cell_type": "code",
763
   "execution_count": null,
764
   "metadata": {},
765
   "outputs": [],
766
   "source": []
767
  },
768
  {
769
   "cell_type": "code",
770
   "execution_count": null,
771
   "metadata": {
772
    "scrolled": true
773
   },
774
   "outputs": [],
775
   "source": []
776
  },
777
  {
778
   "cell_type": "code",
779
   "execution_count": null,
780
   "metadata": {},
781
   "outputs": [],
782
   "source": []
783
  },
784
  {
785
   "cell_type": "code",
786
   "execution_count": 18,
787
   "metadata": {},
788
   "outputs": [
789
    {
790
     "data": {
791
      "image/png": "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\n",
792
      "text/plain": [
793
       "<Figure size 1080x864 with 4 Axes>"
794
      ]
795
     },
796
     "metadata": {
797
      "needs_background": "light"
798
     },
799
     "output_type": "display_data"
800
    }
801
   ],
802
   "source": [
803
    "plt.figure(figsize=(15,12))\n",
804
    "plt.subplot (2,2,1); plt.hist(means,bins=100); plt.title('means')\n",
805
    "plt.subplot (2,2,2); plt.hist(stds, bins=100); plt.title('stds')\n",
806
    "plt.subplot (2,2,3); plt.hist(scale_all,bins=100); plt.title('scale')\n",
807
    "plt.subplot (224);   plt.scatter(shape_in_all[:,0],shape_in_all[:,1],); plt.title('shapes')\n",
808
    "plt.show()"
809
   ]
810
  },
811
  {
812
   "cell_type": "code",
813
   "execution_count": 25,
814
   "metadata": {},
815
   "outputs": [
816
    {
817
     "data": {
818
      "image/png": "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\n",
819
      "text/plain": [
820
       "<Figure size 1080x576 with 2 Axes>"
821
      ]
822
     },
823
     "metadata": {
824
      "needs_background": "light"
825
     },
826
     "output_type": "display_data"
827
    }
828
   ],
829
   "source": [
830
    "plt.figure(figsize=(15,8))\n",
831
    "plt.subplot(121); plt.hist(xtrain.mean(axis=(1,2,3)),bins=100); plt.grid()\n",
832
    "plt.subplot(122); plt.hist(xtrain.std(axis=(1,2,3)),bins=100)\n",
833
    "plt.show()"
834
   ]
835
  },
836
  {
837
   "cell_type": "code",
838
   "execution_count": null,
839
   "metadata": {},
840
   "outputs": [],
841
   "source": []
842
  },
843
  {
844
   "cell_type": "code",
845
   "execution_count": 21,
846
   "metadata": {},
847
   "outputs": [],
848
   "source": [
849
    "def CLAHEF (image_path, two=False, **kwarg) :\n",
850
    "    rgb = cv2.imread(image_path)\n",
851
    "    if two : return(rgb,CLAHEA(rgb,**kwarg))\n",
852
    "    return(CLAHEA(rgb,**kwarg))\n",
853
    "\n",
854
    "def CLAHEA (rgb, gridsize=(94,77)) :\n",
855
    "    lab = cv2.cvtColor(rgb, cv2.COLOR_RGB2LAB)\n",
856
    "    lab_planes = cv2.split(lab)\n",
857
    "    clahe = cv2.createCLAHE(clipLimit=2.0,tileGridSize=gridsize) #(gridsize,gridsize))\n",
858
    "    lab_planes[0] = clahe.apply(lab_planes[0])\n",
859
    "    lab = cv2.merge(lab_planes)\n",
860
    "    rgb1= cv2.cvtColor(lab, cv2.COLOR_LAB2RGB)\n",
861
    "    return(rgb1)\n",
862
    "\n",
863
    "def CLAHEAGray (gray, gridsize=(94,77)) :\n",
864
    "    clahe = cv2.createCLAHE(clipLimit=1.0,tileGridSize=gridsize) #(gridsize,gridsize))\n",
865
    "    gray1 = clahe.apply(gray)\n",
866
    "    return(gray1)\n",
867
    "\n",
868
    "def xCreate2F (nameFile, shape_net, bwOK=True, scale=0.3, \n",
869
    "               contrast=True, clahe=False, count=None) :\n",
870
    "    \n",
871
    "    dd = (1,) if bwOK else (3,)\n",
872
    "    bw = 0    if bwOK else -1\n",
873
    "    \n",
874
    "    zx        = np.zeros(shape_net, dtype=np.float32)\n",
875
    "    \n",
876
    "    img       = cv2.imread(nameFile,bw)\n",
877
    "    shape_in  = img.shape[:2]\n",
878
    "    \n",
879
    "    if clahe and img.ndim==2 : img = CLAHEAGray(img)\n",
880
    "    \n",
881
    "    if type(scale)==type(2.6) or type(scale)==type(2) : # scale_ is float \n",
882
    "        scale_  = scale\n",
883
    "        newsize=(int(scale_*shape_in[1]),int(scale_*shape_in[0]))\n",
884
    "        newimg    = cv2.resize(img,newsize)\n",
885
    "    else :                                              # scale_ is list or tuple (rows,columns) - new size\n",
886
    "        scale_    = min(scale[0]/shape_in[0],scale[1]/shape_in[1])\n",
887
    "        newsize   = (int(shape_in[1]*scale_),int(shape_in[0]*scale_))\n",
888
    "        #print('????',scale,scale_,shape_in,newsize)\n",
889
    "        newimg    = cv2.resize(img,newsize)\n",
890
    "        \n",
891
    "    if 1 :\n",
892
    "        newimg    = newimg.astype(float)\n",
893
    "        newimg    = newimg-newimg.mean()\n",
894
    "        if contrast : newimg    = newimg/(np.abs(newimg)).max()\n",
895
    "    if 1 :\n",
896
    "        newimg    = np.clip((newimg*127+127),0,255).astype(np.uint8)\n",
897
    "    \n",
898
    "    if bwOK : newimg = newimg.reshape(newimg.shape+dd)\n",
899
    "\n",
900
    "    deltas, x = [], []\n",
901
    "    \n",
902
    "    zx        = np.zeros(shape_net, dtype=newimg.dtype)\n",
903
    "    \n",
904
    "    #print(newimg.shape,shape_net)\n",
905
    "    \n",
906
    "    for rr in range(0,newimg.shape[0],shape_net[0]) :\n",
907
    "        for cc in range(0,newimg.shape[1],shape_net[1]) :\n",
908
    "            xx   = zx.copy()\n",
909
    "            ximg = newimg[rr:rr+shape_net[0],cc:cc+shape_net[1]] \n",
910
    "            \n",
911
    "            xx[0:ximg.shape[0],0:ximg.shape[1]] = ximg.copy()\n",
912
    "            \n",
913
    "            if rr>0 or cc>0 :\n",
914
    "                if count is not None and count==0 and ximg.size <= 0.75*zx.size : continue\n",
915
    "                if count is not None and ximg.size <= 0.5*zx.size : continue\n",
916
    "            \n",
917
    "                if (xx==255).all() : continue\n",
918
    "                if (xx==127).all() : continue\n",
919
    "                if (xx==0).all()   : continue\n",
920
    "            \n",
921
    "            #xx   = xx.reshape((1,)+xx.shape)\n",
922
    "            x.append(xx.copy())\n",
923
    "            deltas.append([rr,cc])\n",
924
    "            \n",
925
    "            #plt.imshow(xx.squeeze());\n",
926
    "            #plt.show()\n",
927
    "     \n",
928
    "    x = np.stack(x,axis=0)\n",
929
    "    #x = np.stack(x)\n",
930
    "            \n",
931
    "    return(x,deltas,scale_,shape_in,img.mean(),img.std())\n",
932
    "\n",
933
    "def xCreate2FF(dirData, nFiles_, shape_net, ext='.jpg', counts=None, **kwarg) :\n",
934
    "    nFiles = nFiles_\n",
935
    "    if ext is not None : nFiles = [nFile+ext for nFile in nFiles]\n",
936
    "        \n",
937
    "    aCount, count = 0, None\n",
938
    "    for ii,nFile in enumerate(nFiles) :\n",
939
    "        if counts is not None : count = counts[ii]\n",
940
    "        xdata, deltas, scale, shape_in, xmean, xstd = xCreate2F(os.path.join(dirData,nFile), \n",
941
    "                                                                shape_net, count=count, **kwarg)\n",
942
    "        aCount = aCount + len(xdata)\n",
943
    "        \n",
944
    "    for ii,nFile in enumerate(nFiles) :\n",
945
    "        if counts is not None : count = counts[ii]\n",
946
    "        xdata, xdelta, scale, shape_in, xmean, xstd = xCreate2F(os.path.join(dirData,nFile), \n",
947
    "                                                                shape_net, count=count, **kwarg)\n",
948
    "        if ii==0 :\n",
949
    "            xdatas  = np.zeros((aCount,)+xdata[0].shape, dtype=xdata.dtype)\n",
950
    "            xdeltas = []\n",
951
    "            xscale  = np.zeros((len(nFiles),), dtype=float)\n",
952
    "            xshape  = np.zeros((len(nFiles),len(shape_in)), dtype=int)\n",
953
    "            iiCount = 0\n",
954
    "        xdeltas.append(xdelta)\n",
955
    "        xdatas[iiCount:iiCount+len(xdata)], xscale[ii], xshape[ii]  = xdata.copy(), scale, shape_in\n",
956
    "        iiCount = iiCount+len(xdata)\n",
957
    "        \n",
958
    "    return xdatas,xdeltas,xscale,xshape"
959
   ]
960
  },
961
  {
962
   "cell_type": "code",
963
   "execution_count": null,
964
   "metadata": {},
965
   "outputs": [],
966
   "source": []
967
  },
968
  {
969
   "cell_type": "code",
970
   "execution_count": 22,
971
   "metadata": {},
972
   "outputs": [],
973
   "source": [
974
    "shape_net = (1024,1024+256,1)\n",
975
    "nameFiles = trainYY.image_id.values\n",
976
    "scale  = 0.6\n",
977
    "scale  = (2*1024,1024+256)\n",
978
    "counts = [len(np.array(ll.split()).reshape(-1,5)) for ll in trainYY.fillna('').labels]"
979
   ]
980
  },
981
  {
982
   "cell_type": "code",
983
   "execution_count": 23,
984
   "metadata": {
985
    "scrolled": false
986
   },
987
   "outputs": [],
988
   "source": [
989
    "if 0 :\n",
990
    "    print(datetime.datetime.now())\n",
991
    "    try :    del xtrain; \n",
992
    "    except : pass\n",
993
    "    try :    shape_in_all; \n",
994
    "    except : pass\n",
995
    "    \n",
996
    "    xtrain, xdeltas, scale_all, shape_in_all = xCreate2FF(dirTrain, nameFiles, shape_net, \n",
997
    "                                                          scale=scale, counts=counts, \n",
998
    "                                                          contrast=False, clahe=True)\n",
999
    "    print(datetime.datetime.now())\n",
1000
    "    \n",
1001
    "    print(xtrain.shape, xtrain.dtype, scale_all.shape, shape_in_all.shape, \n",
1002
    "          '\\n', scale_all[:5].tolist(), '\\n',\n",
1003
    "          shape_in_all[:5].tolist())"
1004
   ]
1005
  },
1006
  {
1007
   "cell_type": "code",
1008
   "execution_count": null,
1009
   "metadata": {},
1010
   "outputs": [],
1011
   "source": []
1012
  },
1013
  {
1014
   "cell_type": "code",
1015
   "execution_count": 24,
1016
   "metadata": {},
1017
   "outputs": [
1018
    {
1019
     "data": {
1020
      "image/png": "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\n",
1021
      "text/plain": [
1022
       "<Figure size 432x288 with 1 Axes>"
1023
      ]
1024
     },
1025
     "metadata": {
1026
      "needs_background": "light"
1027
     },
1028
     "output_type": "display_data"
1029
    }
1030
   ],
1031
   "source": [
1032
    "plt.hist(scale_all,bins=100); plt.show()"
1033
   ]
1034
  },
1035
  {
1036
   "cell_type": "code",
1037
   "execution_count": 25,
1038
   "metadata": {},
1039
   "outputs": [],
1040
   "source": [
1041
    "trainYYL = []\n",
1042
    "for ii,ll in enumerate(trainYY.fillna('').labels[:len(xdeltas)]) :\n",
1043
    "    if ll=='' : \n",
1044
    "        trainYYL = trainYYL + ['']*len(xdeltas[ii])\n",
1045
    "        continue\n",
1046
    "    bb         = np.array(ll.split()).reshape((-1,5))\n",
1047
    "    bb1        = bb[:,1:5].astype(float)\n",
1048
    "    bb1[:,2:4] = (bb1[:,2:4]+bb1[:,0:2])\n",
1049
    "    bb1        = bb1*scale_all[ii]\n",
1050
    "    bbc        = (bb1[:,0:2]+bb1[:,2:4])/2\n",
1051
    "    #print(ii,len(bb),xdeltas[ii])\n",
1052
    "    for dd in xdeltas[ii] :\n",
1053
    "        ddx = np.array([dd[1],dd[0],dd[1]+shape_net[1],dd[0]+shape_net[0]])\n",
1054
    "        bbx = ((ddx[0]<=bbc[:,0]) & (ddx[1]<=bbc[:,1]) & (bbc[:,0]<=ddx[2]) & (bbc[:,1]<=ddx[3]))\n",
1055
    "        #print('--',ii,dd,len(bbx[bbx]))\n",
1056
    "        if not bbx.any() : label = ''\n",
1057
    "        else             : \n",
1058
    "            bb2 = (bb1[bbx,0:4]-(dd[1],dd[0],dd[1],dd[0]))\n",
1059
    "            bb2[:,2:4] = bb2[:,2:4]-bb2[:,0:2]\n",
1060
    "            bb2 = bb2.astype(int).astype(str)\n",
1061
    "            #print(bb2.shape,bb.shape,bb[bbx].shape)\n",
1062
    "            bb2 = np.hstack([bb[bbx,:1],bb2])\n",
1063
    "            label = ' '.join([' '.join(bbi) for bbi in bb2.tolist()])\n",
1064
    "        trainYYL.append(label)\n",
1065
    "    #if ii==10 : break\n",
1066
    "trainYYA = np.array(trainYYL)"
1067
   ]
1068
  },
1069
  {
1070
   "cell_type": "code",
1071
   "execution_count": null,
1072
   "metadata": {
1073
    "scrolled": false
1074
   },
1075
   "outputs": [],
1076
   "source": []
1077
  },
1078
  {
1079
   "cell_type": "code",
1080
   "execution_count": 26,
1081
   "metadata": {},
1082
   "outputs": [
1083
    {
1084
     "data": {
1085
      "text/plain": [
1086
       "(7694, 3881, (7694, 1024, 1280, 1))"
1087
      ]
1088
     },
1089
     "execution_count": 26,
1090
     "metadata": {},
1091
     "output_type": "execute_result"
1092
    }
1093
   ],
1094
   "source": [
1095
    "len(trainYYL), len(shape_in_all), xtrain.shape"
1096
   ]
1097
  },
1098
  {
1099
   "cell_type": "code",
1100
   "execution_count": 27,
1101
   "metadata": {},
1102
   "outputs": [
1103
    {
1104
     "data": {
1105
      "text/plain": [
1106
       "array(['hnsd012-039', '100249537_00065_1', '100249416_00027_1'],\n",
1107
       "      dtype=object)"
1108
      ]
1109
     },
1110
     "execution_count": 27,
1111
     "metadata": {},
1112
     "output_type": "execute_result"
1113
    }
1114
   ],
1115
   "source": [
1116
    "nameFiles[:3]"
1117
   ]
1118
  },
1119
  {
1120
   "cell_type": "code",
1121
   "execution_count": null,
1122
   "metadata": {},
1123
   "outputs": [],
1124
   "source": []
1125
  },
1126
  {
1127
   "cell_type": "code",
1128
   "execution_count": null,
1129
   "metadata": {},
1130
   "outputs": [],
1131
   "source": []
1132
  },
1133
  {
1134
   "cell_type": "code",
1135
   "execution_count": 28,
1136
   "metadata": {
1137
    "scrolled": false
1138
   },
1139
   "outputs": [
1140
    {
1141
     "data": {
1142
      "image/png": "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\n",
1143
      "text/plain": [
1144
       "<Figure size 720x1080 with 1 Axes>"
1145
      ]
1146
     },
1147
     "metadata": {
1148
      "needs_background": "light"
1149
     },
1150
     "output_type": "display_data"
1151
    },
1152
    {
1153
     "name": "stdout",
1154
     "output_type": "stream",
1155
     "text": [
1156
      "106\n"
1157
     ]
1158
    },
1159
    {
1160
     "data": {
1161
      "image/png": "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\n",
1162
      "text/plain": [
1163
       "<Figure size 720x1080 with 1 Axes>"
1164
      ]
1165
     },
1166
     "metadata": {
1167
      "needs_background": "light"
1168
     },
1169
     "output_type": "display_data"
1170
    },
1171
    {
1172
     "name": "stdout",
1173
     "output_type": "stream",
1174
     "text": [
1175
      "17\n"
1176
     ]
1177
    },
1178
    {
1179
     "data": {
1180
      "image/png": "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\n",
1181
      "text/plain": [
1182
       "<Figure size 720x1080 with 1 Axes>"
1183
      ]
1184
     },
1185
     "metadata": {
1186
      "needs_background": "light"
1187
     },
1188
     "output_type": "display_data"
1189
    },
1190
    {
1191
     "name": "stdout",
1192
     "output_type": "stream",
1193
     "text": [
1194
      "0\n"
1195
     ]
1196
    },
1197
    {
1198
     "data": {
1199
      "image/png": "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\n",
1200
      "text/plain": [
1201
       "<Figure size 720x1080 with 1 Axes>"
1202
      ]
1203
     },
1204
     "metadata": {
1205
      "needs_background": "light"
1206
     },
1207
     "output_type": "display_data"
1208
    },
1209
    {
1210
     "name": "stdout",
1211
     "output_type": "stream",
1212
     "text": [
1213
      "70\n"
1214
     ]
1215
    },
1216
    {
1217
     "data": {
1218
      "image/png": "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\n",
1219
      "text/plain": [
1220
       "<Figure size 720x1080 with 1 Axes>"
1221
      ]
1222
     },
1223
     "metadata": {
1224
      "needs_background": "light"
1225
     },
1226
     "output_type": "display_data"
1227
    },
1228
    {
1229
     "name": "stdout",
1230
     "output_type": "stream",
1231
     "text": [
1232
      "77\n"
1233
     ]
1234
    }
1235
   ],
1236
   "source": [
1237
    "#plt.figure(figsize=(10,15))\n",
1238
    "#plt.imshow(xtrain[120].squeeze().astype(np.float32));plt.show()\n",
1239
    "xtrain.nbytes/5*3881/(1024*1024*1024)\n",
1240
    "for ii in range(540,545) :\n",
1241
    "#for ii in range(0,10) :\n",
1242
    "    drawSub1A(xtrain[ii],trainYYL[ii],centerXY=False, iiNumber=ii)"
1243
   ]
1244
  },
1245
  {
1246
   "cell_type": "code",
1247
   "execution_count": null,
1248
   "metadata": {},
1249
   "outputs": [],
1250
   "source": []
1251
  },
1252
  {
1253
   "cell_type": "code",
1254
   "execution_count": null,
1255
   "metadata": {},
1256
   "outputs": [],
1257
   "source": []
1258
  },
1259
  {
1260
   "cell_type": "code",
1261
   "execution_count": null,
1262
   "metadata": {},
1263
   "outputs": [],
1264
   "source": []
1265
  },
1266
  {
1267
   "cell_type": "code",
1268
   "execution_count": 28,
1269
   "metadata": {},
1270
   "outputs": [
1271
    {
1272
     "data": {
1273
      "text/plain": [
1274
       "(683464, 2)"
1275
      ]
1276
     },
1277
     "execution_count": 28,
1278
     "metadata": {},
1279
     "output_type": "execute_result"
1280
    }
1281
   ],
1282
   "source": [
1283
    "#X  = [(np.array(ll.split()).reshape((-1,5)))[:,3:].astype(int) for ii,ll in enumerate(trainYY.fillna('').labels)]\n",
1284
    "X  = [(np.array(ll.split()).reshape((-1,5)))[:,3:].astype(int) for ii,ll in enumerate(trainYY.fillna('').labels)]\n",
1285
    "X  = np.vstack([wh*scale_all[ii] for ii,wh in enumerate(X)])\n",
1286
    "X.shape"
1287
   ]
1288
  },
1289
  {
1290
   "cell_type": "code",
1291
   "execution_count": 29,
1292
   "metadata": {},
1293
   "outputs": [
1294
    {
1295
     "data": {
1296
      "text/plain": [
1297
       "(array([[60.25113147, 73.77496765],\n",
1298
       "        [37.65402499, 46.50537948]]),\n",
1299
       " array([1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 0,\n",
1300
       "        0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0,\n",
1301
       "        0, 0, 0, 0, 1, 0], dtype=int32))"
1302
      ]
1303
     },
1304
     "execution_count": 29,
1305
     "metadata": {},
1306
     "output_type": "execute_result"
1307
    }
1308
   ],
1309
   "source": [
1310
    "from sklearn.cluster import KMeans\n",
1311
    "kmeans = KMeans(n_clusters=n_anchors, random_state=0).fit(X)\n",
1312
    "#kmeans.labels_\n",
1313
    "kmeans.cluster_centers_, kmeans.labels_[:50]"
1314
   ]
1315
  },
1316
  {
1317
   "cell_type": "code",
1318
   "execution_count": null,
1319
   "metadata": {},
1320
   "outputs": [],
1321
   "source": []
1322
  },
1323
  {
1324
   "cell_type": "code",
1325
   "execution_count": 30,
1326
   "metadata": {},
1327
   "outputs": [
1328
    {
1329
     "data": {
1330
      "text/plain": [
1331
       "[[[18.82701249271699, 23.252689737692357]],\n",
1332
       " [[90.37669721074036, 110.66245146836516]]]"
1333
      ]
1334
     },
1335
     "execution_count": 30,
1336
     "metadata": {},
1337
     "output_type": "execute_result"
1338
    }
1339
   ],
1340
   "source": [
1341
    "anchors = kmeans.cluster_centers_.astype(np.float64);\n",
1342
    "anchors = anchors[(anchors[:,0]*anchors[:,1]).argsort(),:]\n",
1343
    "##anchors[:int(len(anchors)/2)] *= 0.5\n",
1344
    "##anchors[int(len(anchors)/2):] *= 3.0\n",
1345
    "\n",
1346
    "anchors[0] = anchors[0]/2\n",
1347
    "\n",
1348
    "for ii in range(1,anchors.shape[0]) :\n",
1349
    "    anchors[ii] = anchors[ii]*(1.5**(ii))\n",
1350
    "    \n",
1351
    "#anchors = anchors[(-anchors[:,0]*anchors[:,1]).argsort(),:]\n",
1352
    "###anchors = anchors.astype(np.int32)\n",
1353
    "anchorsl= anchors.copy()\n",
1354
    "anchors = anchors.tolist()\n",
1355
    "anchors = [anchors[:int(len(anchors)/2)],anchors[int(len(anchors)/2):]]\n",
1356
    "anchors_net = anchorsl.astype(np.float64)\n",
1357
    "anchors"
1358
   ]
1359
  },
1360
  {
1361
   "cell_type": "code",
1362
   "execution_count": null,
1363
   "metadata": {},
1364
   "outputs": [],
1365
   "source": []
1366
  },
1367
  {
1368
   "cell_type": "code",
1369
   "execution_count": 31,
1370
   "metadata": {},
1371
   "outputs": [],
1372
   "source": [
1373
    "if 0 :\n",
1374
    "    plt.figure(figsize=(15,10))\n",
1375
    "    plt.scatter(X[:,0],X[:,1],c=kmeans.labels_);\n",
1376
    "    plt.scatter(kmeans.cluster_centers_[:,0],kmeans.cluster_centers_[:,1],c='r',marker='s')\n",
1377
    "    plt.scatter(anchorsl[:,0],anchorsl[:,1],c='r',marker='o')\n",
1378
    "    plt.show()\n",
1379
    "    plt.close('all')"
1380
   ]
1381
  },
1382
  {
1383
   "cell_type": "code",
1384
   "execution_count": null,
1385
   "metadata": {},
1386
   "outputs": [],
1387
   "source": []
1388
  },
1389
  {
1390
   "cell_type": "code",
1391
   "execution_count": 32,
1392
   "metadata": {},
1393
   "outputs": [],
1394
   "source": [
1395
    "if 0 :\n",
1396
    "    print(datetime.datetime.now(), dirSaveX)\n",
1397
    "\n",
1398
    "    if not os.path.exists(dirSaveX) : os.mkdir(dirSaveX)\n",
1399
    "    \n",
1400
    "    np.save(os.path.join(dirSaveX,'xtrain_files.npy'),nameFiles)\n",
1401
    "    np.save(os.path.join(dirSaveX,'xtrain_scale.npy'),scale_all)\n",
1402
    "    np.save(os.path.join(dirSaveX,'xtrain_shape_in.npy'),shape_in_all)\n",
1403
    "\n",
1404
    "    np.save(os.path.join(dirSaveX,'anchors_net.npy'),anchors_net)\n",
1405
    "    np.save(os.path.join(dirSaveX,'xtrain_xdeltas.npy'),xdeltas)\n",
1406
    "\n",
1407
    "    np.save(os.path.join(dirSaveX,'xtrain.npy'),xtrain)\n",
1408
    "\n",
1409
    "\n",
1410
    "    try :    np.save(os.path.join(dirSaveX,'split.npy'),split)\n",
1411
    "    except : abc = 1\n",
1412
    "\n",
1413
    "    print(datetime.datetime.now())"
1414
   ]
1415
  },
1416
  {
1417
   "cell_type": "code",
1418
   "execution_count": null,
1419
   "metadata": {},
1420
   "outputs": [],
1421
   "source": []
1422
  },
1423
  {
1424
   "cell_type": "code",
1425
   "execution_count": null,
1426
   "metadata": {},
1427
   "outputs": [],
1428
   "source": []
1429
  },
1430
  {
1431
   "cell_type": "code",
1432
   "execution_count": 33,
1433
   "metadata": {},
1434
   "outputs": [
1435
    {
1436
     "data": {
1437
      "text/plain": [
1438
       "array([[ 18.82701249,  23.25268974],\n",
1439
       "       [ 90.37669721, 110.66245147]])"
1440
      ]
1441
     },
1442
     "execution_count": 33,
1443
     "metadata": {},
1444
     "output_type": "execute_result"
1445
    }
1446
   ],
1447
   "source": [
1448
    "anchors_net"
1449
   ]
1450
  },
1451
  {
1452
   "cell_type": "code",
1453
   "execution_count": 34,
1454
   "metadata": {},
1455
   "outputs": [],
1456
   "source": [
1457
    "anchorsl= anchors_net.copy()\n",
1458
    "anchors = anchorsl.tolist()\n",
1459
    "anchors = [anchors[:int(len(anchors)/2)],anchors[int(len(anchors)/2):]]"
1460
   ]
1461
  },
1462
  {
1463
   "cell_type": "code",
1464
   "execution_count": null,
1465
   "metadata": {},
1466
   "outputs": [],
1467
   "source": []
1468
  },
1469
  {
1470
   "cell_type": "code",
1471
   "execution_count": null,
1472
   "metadata": {},
1473
   "outputs": [],
1474
   "source": []
1475
  },
1476
  {
1477
   "cell_type": "code",
1478
   "execution_count": 35,
1479
   "metadata": {},
1480
   "outputs": [
1481
    {
1482
     "name": "stderr",
1483
     "output_type": "stream",
1484
     "text": [
1485
      "Using TensorFlow backend.\n"
1486
     ]
1487
    }
1488
   ],
1489
   "source": [
1490
    "from keras.models import Input, Model\n",
1491
    "from keras.layers import Convolution2D, MaxPooling2D, UpSampling2D, Conv2D, Concatenate, Activation, Dropout,Add\n",
1492
    "from keras.layers import Conv2DTranspose, SpatialDropout2D, Dense, Reshape, Flatten, AveragePooling2D\n",
1493
    "from keras.layers.normalization import BatchNormalization\n",
1494
    "from keras.callbacks import ModelCheckpoint, ReduceLROnPlateau, EarlyStopping, TensorBoard\n",
1495
    "import keras\n",
1496
    "import keras.backend as K "
1497
   ]
1498
  },
1499
  {
1500
   "cell_type": "code",
1501
   "execution_count": 36,
1502
   "metadata": {},
1503
   "outputs": [],
1504
   "source": [
1505
    "import tensorflow as tf"
1506
   ]
1507
  },
1508
  {
1509
   "cell_type": "code",
1510
   "execution_count": 37,
1511
   "metadata": {},
1512
   "outputs": [
1513
    {
1514
     "data": {
1515
      "text/plain": [
1516
       "'2.2.4'"
1517
      ]
1518
     },
1519
     "execution_count": 37,
1520
     "metadata": {},
1521
     "output_type": "execute_result"
1522
    }
1523
   ],
1524
   "source": [
1525
    "keras.__version__"
1526
   ]
1527
  },
1528
  {
1529
   "cell_type": "code",
1530
   "execution_count": 38,
1531
   "metadata": {},
1532
   "outputs": [],
1533
   "source": [
1534
    "from keras.models import Input, Model\n",
1535
    "from keras.layers import Convolution2D, MaxPooling2D, UpSampling2D, Conv2D, Concatenate, Activation, Dropout\n",
1536
    "from keras.layers import Conv2DTranspose, RepeatVector, ZeroPadding2D, Cropping2D, multiply, average, add, subtract\n",
1537
    "from keras.layers import GlobalAveragePooling2D, GlobalMaxPooling2D, GlobalAveragePooling1D\n",
1538
    "from keras.layers.normalization import BatchNormalization\n",
1539
    "\n",
1540
    "\n",
1541
    "def se_layer(x, out_dim, ratio = 4) : #4):\n",
1542
    "    '''\n",
1543
    "    SE module performs inter-channel weighting.\n",
1544
    "    '''\n",
1545
    "    squeeze = GlobalAveragePooling2D()(x)\n",
1546
    "\n",
1547
    "    excitation = Dense(units=out_dim // ratio)(squeeze)\n",
1548
    "    excitation = Activation('relu')(excitation)\n",
1549
    "    excitation = Dense(units=out_dim)(excitation)\n",
1550
    "    excitation = Activation('sigmoid')(excitation)\n",
1551
    "    excitation = Reshape((1,1,out_dim))(excitation)\n",
1552
    "\n",
1553
    "    scale = multiply([x,excitation])\n",
1554
    "\n",
1555
    "    return scale\n",
1556
    "\n",
1557
    "def residual_block(blockInput, num_filters=16, seOK = True, batch_activate = False):\n",
1558
    "    x = BatchActivate(blockInput)\n",
1559
    "    x = convolution_block(x, num_filters, (3,3))\n",
1560
    "    x = convolution_block(x, num_filters, (3,3), activation=False)\n",
1561
    "    \n",
1562
    "    if seOK : x = se_layer(x,num_filters)\n",
1563
    "    \n",
1564
    "    x = Add()([x, blockInput])\n",
1565
    "    if batch_activate:\n",
1566
    "        x = BatchActivate(x)\n",
1567
    "    return x\n",
1568
    "\n",
1569
    "def BatchActivate(x):\n",
1570
    "    x = BatchNormalization()(x)\n",
1571
    "    x = Activation('relu')(x)\n",
1572
    "    return x\n",
1573
    "\n",
1574
    "def convolution_block(x, filters, size, strides=(1,1), padding='same', activation=True):\n",
1575
    "    x = Conv2D(filters, size, strides=strides, padding=padding, kernel_initializer='he_normal')(x)\n",
1576
    "    if activation == True:\n",
1577
    "        x = BatchActivate(x)\n",
1578
    "    return x\n"
1579
   ]
1580
  },
1581
  {
1582
   "cell_type": "code",
1583
   "execution_count": 88,
1584
   "metadata": {},
1585
   "outputs": [
1586
    {
1587
     "name": "stdout",
1588
     "output_type": "stream",
1589
     "text": [
1590
      "__________________________________________________________________________________________________\n",
1591
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
1592
      "==================================================================================================\n",
1593
      "input (InputLayer)              (None, 1024, 1280, 1 0                                            \n",
1594
      "__________________________________________________________________________________________________\n",
1595
      "conv2d_27 (Conv2D)              (None, 1024, 1280, 1 160         input[0][0]                      \n",
1596
      "__________________________________________________________________________________________________\n",
1597
      "batch_normalization_23 (BatchNo (None, 1024, 1280, 1 64          conv2d_27[0][0]                  \n",
1598
      "__________________________________________________________________________________________________\n",
1599
      "leaky_re_lu_25 (LeakyReLU)      (None, 1024, 1280, 1 0           batch_normalization_23[0][0]     \n",
1600
      "__________________________________________________________________________________________________\n",
1601
      "max_pooling2d_13 (MaxPooling2D) (None, 512, 640, 16) 0           leaky_re_lu_25[0][0]             \n",
1602
      "__________________________________________________________________________________________________\n",
1603
      "conv2d_28 (Conv2D)              (None, 512, 640, 32) 4640        max_pooling2d_13[0][0]           \n",
1604
      "__________________________________________________________________________________________________\n",
1605
      "batch_normalization_24 (BatchNo (None, 512, 640, 32) 128         conv2d_28[0][0]                  \n",
1606
      "__________________________________________________________________________________________________\n",
1607
      "leaky_re_lu_26 (LeakyReLU)      (None, 512, 640, 32) 0           batch_normalization_24[0][0]     \n",
1608
      "__________________________________________________________________________________________________\n",
1609
      "max_pooling2d_14 (MaxPooling2D) (None, 256, 320, 32) 0           leaky_re_lu_26[0][0]             \n",
1610
      "__________________________________________________________________________________________________\n",
1611
      "conv2d_29 (Conv2D)              (None, 256, 320, 64) 18496       max_pooling2d_14[0][0]           \n",
1612
      "__________________________________________________________________________________________________\n",
1613
      "batch_normalization_25 (BatchNo (None, 256, 320, 64) 256         conv2d_29[0][0]                  \n",
1614
      "__________________________________________________________________________________________________\n",
1615
      "leaky_re_lu_27 (LeakyReLU)      (None, 256, 320, 64) 0           batch_normalization_25[0][0]     \n",
1616
      "__________________________________________________________________________________________________\n",
1617
      "max_pooling2d_15 (MaxPooling2D) (None, 128, 160, 64) 0           leaky_re_lu_27[0][0]             \n",
1618
      "__________________________________________________________________________________________________\n",
1619
      "conv2d_30 (Conv2D)              (None, 128, 160, 128 73856       max_pooling2d_15[0][0]           \n",
1620
      "__________________________________________________________________________________________________\n",
1621
      "batch_normalization_26 (BatchNo (None, 128, 160, 128 512         conv2d_30[0][0]                  \n",
1622
      "__________________________________________________________________________________________________\n",
1623
      "leaky_re_lu_28 (LeakyReLU)      (None, 128, 160, 128 0           batch_normalization_26[0][0]     \n",
1624
      "__________________________________________________________________________________________________\n",
1625
      "max_pooling2d_16 (MaxPooling2D) (None, 64, 80, 128)  0           leaky_re_lu_28[0][0]             \n",
1626
      "__________________________________________________________________________________________________\n",
1627
      "conv2d_31 (Conv2D)              (None, 64, 80, 256)  295168      max_pooling2d_16[0][0]           \n",
1628
      "__________________________________________________________________________________________________\n",
1629
      "batch_normalization_27 (BatchNo (None, 64, 80, 256)  1024        conv2d_31[0][0]                  \n",
1630
      "__________________________________________________________________________________________________\n",
1631
      "leaky_re_lu_29 (LeakyReLU)      (None, 64, 80, 256)  0           batch_normalization_27[0][0]     \n",
1632
      "__________________________________________________________________________________________________\n",
1633
      "max_pooling2d_17 (MaxPooling2D) (None, 32, 40, 256)  0           leaky_re_lu_29[0][0]             \n",
1634
      "__________________________________________________________________________________________________\n",
1635
      "conv2d_32 (Conv2D)              (None, 32, 40, 512)  1180160     max_pooling2d_17[0][0]           \n",
1636
      "__________________________________________________________________________________________________\n",
1637
      "batch_normalization_28 (BatchNo (None, 32, 40, 512)  2048        conv2d_32[0][0]                  \n",
1638
      "__________________________________________________________________________________________________\n",
1639
      "leaky_re_lu_30 (LeakyReLU)      (None, 32, 40, 512)  0           batch_normalization_28[0][0]     \n",
1640
      "__________________________________________________________________________________________________\n",
1641
      "max_pooling2d_18 (MaxPooling2D) (None, 31, 39, 512)  0           leaky_re_lu_30[0][0]             \n",
1642
      "__________________________________________________________________________________________________\n",
1643
      "conv2d_33 (Conv2D)              (None, 31, 39, 1024) 4719616     max_pooling2d_18[0][0]           \n",
1644
      "__________________________________________________________________________________________________\n",
1645
      "batch_normalization_29 (BatchNo (None, 31, 39, 1024) 4096        conv2d_33[0][0]                  \n",
1646
      "__________________________________________________________________________________________________\n",
1647
      "leaky_re_lu_31 (LeakyReLU)      (None, 31, 39, 1024) 0           batch_normalization_29[0][0]     \n",
1648
      "__________________________________________________________________________________________________\n",
1649
      "conv2d_34 (Conv2D)              (None, 31, 39, 512)  524800      leaky_re_lu_31[0][0]             \n",
1650
      "__________________________________________________________________________________________________\n",
1651
      "batch_normalization_30 (BatchNo (None, 31, 39, 512)  2048        conv2d_34[0][0]                  \n",
1652
      "__________________________________________________________________________________________________\n",
1653
      "leaky_re_lu_32 (LeakyReLU)      (None, 31, 39, 512)  0           batch_normalization_30[0][0]     \n",
1654
      "__________________________________________________________________________________________________\n",
1655
      "conv2d_35 (Conv2D)              (None, 31, 39, 1024) 4719616     leaky_re_lu_32[0][0]             \n",
1656
      "__________________________________________________________________________________________________\n",
1657
      "conv2d_38 (Conv2D)              (None, 31, 39, 512)  2359808     leaky_re_lu_32[0][0]             \n",
1658
      "__________________________________________________________________________________________________\n",
1659
      "batch_normalization_31 (BatchNo (None, 31, 39, 1024) 4096        conv2d_35[0][0]                  \n",
1660
      "__________________________________________________________________________________________________\n",
1661
      "batch_normalization_33 (BatchNo (None, 31, 39, 512)  2048        conv2d_38[0][0]                  \n",
1662
      "__________________________________________________________________________________________________\n",
1663
      "leaky_re_lu_33 (LeakyReLU)      (None, 31, 39, 1024) 0           batch_normalization_31[0][0]     \n",
1664
      "__________________________________________________________________________________________________\n",
1665
      "leaky_re_lu_35 (LeakyReLU)      (None, 31, 39, 512)  0           batch_normalization_33[0][0]     \n",
1666
      "__________________________________________________________________________________________________\n",
1667
      "conv2d_36 (Conv2D)              (None, 31, 39, 5)    5125        leaky_re_lu_33[0][0]             \n",
1668
      "__________________________________________________________________________________________________\n",
1669
      "conv2d_39 (Conv2D)              (None, 31, 39, 5)    2565        leaky_re_lu_35[0][0]             \n",
1670
      "__________________________________________________________________________________________________\n",
1671
      "reshape_5 (Reshape)             (None, 31, 39, 1, 5) 0           conv2d_36[0][0]                  \n",
1672
      "__________________________________________________________________________________________________\n",
1673
      "reshape_6 (Reshape)             (None, 31, 39, 1, 5) 0           conv2d_39[0][0]                  \n",
1674
      "__________________________________________________________________________________________________\n",
1675
      "concatenate_6 (Concatenate)     (None, 31, 39, 2, 5) 0           reshape_5[0][0]                  \n",
1676
      "                                                                 reshape_6[0][0]                  \n",
1677
      "__________________________________________________________________________________________________\n",
1678
      "leaky_re_lu_36 (LeakyReLU)      (None, 31, 39, 2, 5) 0           concatenate_6[0][0]              \n",
1679
      "==================================================================================================\n",
1680
      "Total params: 13,920,330\n",
1681
      "Trainable params: 13,912,170\n",
1682
      "Non-trainable params: 8,160\n",
1683
      "__________________________________________________________________________________________________\n"
1684
     ]
1685
    }
1686
   ],
1687
   "source": [
1688
    "from keras.models import Input, Model\n",
1689
    "from keras.layers import Convolution2D, MaxPooling2D, UpSampling2D, Conv2D, Conv3D, Concatenate, Activation, Dropout\n",
1690
    "from keras.layers import Conv2DTranspose, RepeatVector, ZeroPadding2D, Cropping2D, multiply, average, add, subtract\n",
1691
    "from keras.layers import GlobalAveragePooling2D, GlobalMaxPooling2D, GlobalAveragePooling1D\n",
1692
    "from keras.layers.normalization import BatchNormalization\n",
1693
    "\n",
1694
    "def BatchActivate(x, activation='relu'):\n",
1695
    "    x = BatchNormalization()(x)\n",
1696
    "    if  activation :\n",
1697
    "        x = keras.layers.LeakyReLU()(x) if activation=='LeakReLu' else Activation(activation)(x)\n",
1698
    "    return x\n",
1699
    "\n",
1700
    "def myDarkNetTiny (i) :\n",
1701
    "    \n",
1702
    "    #i = Input(shape=img_shape,   name='input');\n",
1703
    "    \n",
1704
    "    a = i;\n",
1705
    "    \n",
1706
    "    filters = 16\n",
1707
    "    \n",
1708
    "    for _ in range(4) : #(4) :\n",
1709
    "        a = convolution_block(a, filters, (3,3), activation=False)\n",
1710
    "        a = BatchActivate(a,activation='LeakReLu')\n",
1711
    "        a = MaxPooling2D(2, strides=2)(a)\n",
1712
    "        filters = filters * 2\n",
1713
    "        \n",
1714
    "    #a = convolution_block(a,  256, (3,3), activation=False)\n",
1715
    "    a = convolution_block(a,  filters, (3,3), activation=False)\n",
1716
    "    a = BatchActivate(a,activation='LeakReLu')\n",
1717
    "    r = a\n",
1718
    "    a = MaxPooling2D(2, strides=2)(a)\n",
1719
    "    filters = filters * 2\n",
1720
    "    \n",
1721
    "    #a = convolution_block(a,  512, (3,3), activation=False)\n",
1722
    "    a = convolution_block(a,  filters, (3,3), activation=False)\n",
1723
    "    a = BatchActivate(a,activation='LeakReLu')\n",
1724
    "    a = MaxPooling2D(2, strides=1)(a)\n",
1725
    "    filters = filters * 2\n",
1726
    "    \n",
1727
    "    #a = convolution_block(a, 1024, (3,3), activation=False)\n",
1728
    "    a = convolution_block(a, filters, (3,3), activation=False)\n",
1729
    "    a = BatchActivate(a,activation='LeakReLu')\n",
1730
    "    \n",
1731
    "    return a, r\n",
1732
    "        \n",
1733
    "def feature_pyramid_tiny(i):\n",
1734
    "    \n",
1735
    "    a = i\n",
1736
    "    #a = convolution_block(a,  256, (1,1), activation=False)\n",
1737
    "    a = convolution_block(a,  512, (1,1), activation=False)\n",
1738
    "    a = BatchActivate(a,activation='LeakReLu')\n",
1739
    "    r = a\n",
1740
    "    \n",
1741
    "    #a = convolution_block(a,  512, (3,3), activation=False)\n",
1742
    "    a = convolution_block(a,  1024, (3,3), activation=False)\n",
1743
    "    a = BatchActivate(a,activation='LeakReLu')\n",
1744
    "    \n",
1745
    "    return a, r\n",
1746
    "\n",
1747
    "def yolo_layer(i, n_classes, anchors, img_size):\n",
1748
    "\n",
1749
    "    a = i\n",
1750
    "    n_anchors = len(anchors)\n",
1751
    "    a = convolution_block(a, n_anchors * (5 + n_classes), (1,1), activation=False)\n",
1752
    "\n",
1753
    "    a_shape = K.int_shape(a)[1:-1]+(n_anchors,5+n_classes)\n",
1754
    "    a = keras.layers.Reshape(a_shape)(a)\n",
1755
    "    \n",
1756
    "    return a\n",
1757
    "    return inputs\n",
1758
    "\n",
1759
    "def myYoloV3Tiny (img_shape, num_classes, anchors) :\n",
1760
    "    \n",
1761
    "    i = Input(shape=img_shape,   name='input');\n",
1762
    "    \n",
1763
    "    a = i;\n",
1764
    "    \n",
1765
    "    a, rD = myDarkNetTiny (a)\n",
1766
    "    \n",
1767
    "    a, rP = feature_pyramid_tiny(a)\n",
1768
    "    \n",
1769
    "    d1    = yolo_layer(a, num_classes, anchors[0], img_shape)\n",
1770
    "    \n",
1771
    "    #a     = convolution_block(rP, 128, (1,1), activation=False)\n",
1772
    "    a     = convolution_block(rP, 256, (1,1), activation=False)\n",
1773
    "    a     = BatchActivate(a,activation='LeakReLu')\n",
1774
    "    \n",
1775
    "    a     = UpSampling2D(name='UpSampling')(a)\n",
1776
    "    a     = ZeroPadding2D((1,1))(a)\n",
1777
    "    a     = Concatenate()([a,rD])\n",
1778
    "    \n",
1779
    "    #a     = convolution_block(rP, 256, (3,3), activation=False)\n",
1780
    "    a     = convolution_block(rP, 512, (3,3), activation=False)\n",
1781
    "    a     = BatchActivate(a,activation='LeakReLu')\n",
1782
    "    \n",
1783
    "    d2    = yolo_layer(a, num_classes, anchors[1], img_shape)\n",
1784
    "    \n",
1785
    "    o     = Concatenate(axis=-2)([d1,d2])\n",
1786
    "    o     = keras.layers.LeakyReLU()(o)    \n",
1787
    "    ######o     = Activation('sigmoid')(o)\n",
1788
    "                                            \n",
1789
    "    return Model(inputs=i, outputs=o, name='my-YoloV3-Tiny')\n",
1790
    "\n",
1791
    "if 1 :\n",
1792
    "    \n",
1793
    "    model10   = myYoloV3Tiny(xtrain[0].shape, num_classes=len(classes), anchors=anchors)\n",
1794
    "    \n",
1795
    "    #model10   = myYoloV3Tiny((1024,768,1), num_classes=len(classes), anchors=anchors)\n",
1796
    "    #model10   = mySoloV3Tiny((1024+1024,1024+512,1), num_classes=len(classes_all), anchors=anchors)\n",
1797
    "    model10.summary()    "
1798
   ]
1799
  },
1800
  {
1801
   "cell_type": "code",
1802
   "execution_count": 89,
1803
   "metadata": {},
1804
   "outputs": [
1805
    {
1806
     "data": {
1807
      "text/plain": [
1808
       "(43, (1024, 1280, 1), (31, 39, 2, 5))"
1809
      ]
1810
     },
1811
     "execution_count": 89,
1812
     "metadata": {},
1813
     "output_type": "execute_result"
1814
    }
1815
   ],
1816
   "source": [
1817
    "model10.get_layer(index=0).input_shape\n",
1818
    "len(model10.layers),model10.get_layer(index=0).output_shape[1:], model10.get_layer(index=-1).output_shape[1:]"
1819
   ]
1820
  },
1821
  {
1822
   "cell_type": "code",
1823
   "execution_count": 90,
1824
   "metadata": {},
1825
   "outputs": [
1826
    {
1827
     "data": {
1828
      "text/plain": [
1829
       "0"
1830
      ]
1831
     },
1832
     "execution_count": 90,
1833
     "metadata": {},
1834
     "output_type": "execute_result"
1835
    }
1836
   ],
1837
   "source": [
1838
    "len(classes)"
1839
   ]
1840
  },
1841
  {
1842
   "cell_type": "code",
1843
   "execution_count": null,
1844
   "metadata": {},
1845
   "outputs": [],
1846
   "source": []
1847
  },
1848
  {
1849
   "cell_type": "code",
1850
   "execution_count": null,
1851
   "metadata": {},
1852
   "outputs": [],
1853
   "source": []
1854
  },
1855
  {
1856
   "cell_type": "code",
1857
   "execution_count": 91,
1858
   "metadata": {},
1859
   "outputs": [
1860
    {
1861
     "data": {
1862
      "text/plain": [
1863
       "[[18.82701249271699, 23.252689737692357],\n",
1864
       " [90.37669721074036, 110.66245146836516]]"
1865
      ]
1866
     },
1867
     "execution_count": 91,
1868
     "metadata": {},
1869
     "output_type": "execute_result"
1870
    }
1871
   ],
1872
   "source": [
1873
    "anchorsl.tolist()"
1874
   ]
1875
  },
1876
  {
1877
   "cell_type": "code",
1878
   "execution_count": null,
1879
   "metadata": {},
1880
   "outputs": [],
1881
   "source": []
1882
  },
1883
  {
1884
   "cell_type": "code",
1885
   "execution_count": null,
1886
   "metadata": {},
1887
   "outputs": [],
1888
   "source": []
1889
  },
1890
  {
1891
   "cell_type": "code",
1892
   "execution_count": 92,
1893
   "metadata": {},
1894
   "outputs": [],
1895
   "source": [
1896
    "batch_size=4\n",
1897
    "#batch_size=3\n",
1898
    "#batch_size=2\n",
1899
    "batch_size=32\n",
1900
    "\n",
1901
    "\n",
1902
    "shape_out   = np.array(model10.get_layer(index=-1).output_shape[1:]);\n",
1903
    "grid_net    = shape_net[0:2]/shape_out[0:2]\n",
1904
    "#scale_mean  = 0.32114449904722703\n",
1905
    "#anchors_net = (anchorsl*scale_mean).astype(np.float64)"
1906
   ]
1907
  },
1908
  {
1909
   "cell_type": "code",
1910
   "execution_count": 93,
1911
   "metadata": {},
1912
   "outputs": [],
1913
   "source": [
1914
    "#classes = classes_all[0:500]; classes[:10]"
1915
   ]
1916
  },
1917
  {
1918
   "cell_type": "code",
1919
   "execution_count": 94,
1920
   "metadata": {},
1921
   "outputs": [
1922
    {
1923
     "data": {
1924
      "text/plain": [
1925
       "([[18.82701249271699, 23.252689737692357],\n",
1926
       "  [90.37669721074036, 110.66245146836516]],\n",
1927
       " array([33.03225806, 32.82051282]))"
1928
      ]
1929
     },
1930
     "execution_count": 94,
1931
     "metadata": {},
1932
     "output_type": "execute_result"
1933
    }
1934
   ],
1935
   "source": [
1936
    "anchors_net.tolist(), grid_net"
1937
   ]
1938
  },
1939
  {
1940
   "cell_type": "code",
1941
   "execution_count": null,
1942
   "metadata": {},
1943
   "outputs": [],
1944
   "source": []
1945
  },
1946
  {
1947
   "cell_type": "code",
1948
   "execution_count": 95,
1949
   "metadata": {
1950
    "scrolled": false
1951
   },
1952
   "outputs": [],
1953
   "source": [
1954
    "if 0 :\n",
1955
    "    print(datetime.datetime.now())\n",
1956
    "\n",
1957
    "\n",
1958
    "    print('net shape:', shape_net)\n",
1959
    "    print('out shape:', shape_out)\n",
1960
    "    print('net grid:',  grid_net)\n",
1961
    "\n",
1962
    "    print('n_anchors:', len(anchorsl))\n",
1963
    "    print('anchors:', anchorsl.tolist())\n",
1964
    "\n",
1965
    "    scale_mean = scale_all.mean()\n",
1966
    "\n",
1967
    "    print('\\nscale_mean=',scale_mean,'\\nanchors_net=',anchors_net.tolist())\n",
1968
    "\n",
1969
    "\n",
1970
    "    try    : del ytrain\n",
1971
    "    except : abc = 1\n",
1972
    "\n",
1973
    "\n",
1974
    "    ytrain, ccc = yTrainCreate (range(len(trainYY)), trainYY, classes,\n",
1975
    "                           shape_net, shape_out, anchors_net, \n",
1976
    "                           scale_all, shape_in_all, printOK=True)\n",
1977
    "\n",
1978
    "    print('ccc errors',ccc)\n",
1979
    "\n",
1980
    "    print('\\n',datetime.datetime.now())\n",
1981
    "    np.round(xtrain.nbytes/(1024*1024*1024),3), xtrain.shape, ytrain.shape, np.round(ytrain.nbytes/(1024*1024*1024),3)"
1982
   ]
1983
  },
1984
  {
1985
   "cell_type": "code",
1986
   "execution_count": null,
1987
   "metadata": {},
1988
   "outputs": [],
1989
   "source": []
1990
  },
1991
  {
1992
   "cell_type": "code",
1993
   "execution_count": 96,
1994
   "metadata": {},
1995
   "outputs": [
1996
    {
1997
     "data": {
1998
      "text/plain": [
1999
       "(9.392, (7694, 1024, 1280, 1))"
2000
      ]
2001
     },
2002
     "execution_count": 96,
2003
     "metadata": {},
2004
     "output_type": "execute_result"
2005
    }
2006
   ],
2007
   "source": [
2008
    "np.round(xtrain.nbytes/(1024*1024*1024),3), xtrain.shape"
2009
   ]
2010
  },
2011
  {
2012
   "cell_type": "code",
2013
   "execution_count": 97,
2014
   "metadata": {},
2015
   "outputs": [
2016
    {
2017
     "data": {
2018
      "text/plain": [
2019
       "[]"
2020
      ]
2021
     },
2022
     "execution_count": 97,
2023
     "metadata": {},
2024
     "output_type": "execute_result"
2025
    }
2026
   ],
2027
   "source": [
2028
    "classes[:20]"
2029
   ]
2030
  },
2031
  {
2032
   "cell_type": "code",
2033
   "execution_count": 98,
2034
   "metadata": {},
2035
   "outputs": [
2036
    {
2037
     "data": {
2038
      "text/plain": [
2039
       "array([0., 0., 0., 0., 0.])"
2040
      ]
2041
     },
2042
     "execution_count": 98,
2043
     "metadata": {},
2044
     "output_type": "execute_result"
2045
    }
2046
   ],
2047
   "source": [
2048
    "np.sum(np.zeros((5,10,3)),axis=(1,2))"
2049
   ]
2050
  },
2051
  {
2052
   "cell_type": "code",
2053
   "execution_count": 99,
2054
   "metadata": {},
2055
   "outputs": [
2056
    {
2057
     "data": {
2058
      "text/plain": [
2059
       "(array([33.03225806, 32.82051282]), (1024, 1280, 1))"
2060
      ]
2061
     },
2062
     "execution_count": 99,
2063
     "metadata": {},
2064
     "output_type": "execute_result"
2065
    }
2066
   ],
2067
   "source": [
2068
    "grid_net, shape_net"
2069
   ]
2070
  },
2071
  {
2072
   "cell_type": "code",
2073
   "execution_count": 181,
2074
   "metadata": {},
2075
   "outputs": [],
2076
   "source": [
2077
    "batch_size =  3  # (1024+512)x1024\n",
2078
    "batch_size =  5  # 1024x(512+256)\n",
2079
    "batch_size =  4  # 1024x(512+256)\n",
2080
    "batch_size =  2  # 1024x(512+256) v2\n",
2081
    "#batch_size =  1  # 1024x(512+256) v2\n",
2082
    "batch_size =  3  # 1024x1280 v2\n",
2083
    "batch_size =  4  # 1024x1280 v2\n",
2084
    "#batch_size =  5  # 1024x1280 v2"
2085
   ]
2086
  },
2087
  {
2088
   "cell_type": "code",
2089
   "execution_count": null,
2090
   "metadata": {},
2091
   "outputs": [],
2092
   "source": []
2093
  },
2094
  {
2095
   "cell_type": "code",
2096
   "execution_count": 182,
2097
   "metadata": {},
2098
   "outputs": [
2099
    {
2100
     "name": "stdout",
2101
     "output_type": "stream",
2102
     "text": [
2103
      "4 [33.03225806 32.82051282] [31 39  2  5] (4, 31, 39, 2, 2) (4, 31, 39, 2, 2)\n"
2104
     ]
2105
    }
2106
   ],
2107
   "source": [
2108
    "def get_constant1(batch_size,shape_out) :\n",
2109
    "    shape_constant = ([batch_size,]+shape_out[:-1].tolist()+[1,]);\n",
2110
    "    \n",
2111
    "    t4  = np.cumsum(np.ones(shape_constant),axis=2)-1.0\n",
2112
    "    t5  = np.cumsum(np.ones(shape_constant),axis=1)-1.0\n",
2113
    "    \n",
2114
    "    t6  = np.concatenate([t4,t5],axis=-1)\n",
2115
    "    return t6\n",
2116
    "\n",
2117
    "def get_constant2(batch_size,shape_out,anchorsl) :\n",
2118
    "    shape_constant = ([batch_size,]+shape_out[:-1].tolist()+[2,]);\n",
2119
    "    tt  = np.ones(shape_constant, dtype=float)\n",
2120
    "    #print(tt.shape, anchorsl.shape,tt[...,0,:].shape)\n",
2121
    "    for ii, (ww, hh) in enumerate(anchorsl) : \n",
2122
    "        tt[...,ii,:]=np.array([shape_net[1],shape_net[0]])/np.array([ww,hh])\n",
2123
    "    return tt\n",
2124
    "\n",
2125
    "constant1 = get_constant1(batch_size,shape_out).astype(np.float32)\n",
2126
    "constant2 = get_constant2(batch_size,shape_out,anchors_net).astype(np.float32)\n",
2127
    "\n",
2128
    "print(batch_size,grid_net,shape_out,constant1.shape,constant2.shape)\n",
2129
    "\n",
2130
    "def yloss (y_true_, y_pred_, sigmoidOK=True, debug=False) :\n",
2131
    "    \n",
2132
    "    true_conf = y_true_[...,0]\n",
2133
    "    \n",
2134
    "    if shape_out[-1]>5 :\n",
2135
    "        true_class= y_true_[...,5:]\n",
2136
    "    \n",
2137
    "    nbatch    = K.shape(true_conf)[0]\n",
2138
    "    nobjects1 = K.sum(true_conf,axis=(1,2,3))\n",
2139
    "    nobjects0 = K.sum(1.0-true_conf,axis=(1,2,3))\n",
2140
    "    nobjects  = nobjects0+nobjects1\n",
2141
    "    \n",
2142
    "    true_conf1= K.expand_dims(true_conf,axis=-1)\n",
2143
    "    true_conf2= K.concatenate([true_conf1,true_conf1])\n",
2144
    "\n",
2145
    "    pred_conf = y_pred_[...,0]\n",
2146
    "    if sigmoidOK : pred_conf = K.sigmoid(pred_conf)\n",
2147
    "        \n",
2148
    "    #xxx1_conf = K.sum(K.cast(K.greater_equal(pred_conf,0.5),K.floatx()),axis=(1,2,3))\n",
2149
    "    \n",
2150
    "    #pred_conf1= K.expand_dims(pred_conf,axis=-1)\n",
2151
    "    #pred_conf2= K.concatenate([pred_conf1,pred_conf1])\n",
2152
    "    \n",
2153
    "        \n",
2154
    "    conf_loss     = 0.0\n",
2155
    "\n",
2156
    "    if 0 :\n",
2157
    "        conf_loss = conf_loss +  (true_conf*K.binary_crossentropy(true_conf,pred_conf))\n",
2158
    "        conf_loss = conf_loss + ((1-true_conf)*K.binary_crossentropy(true_conf,pred_conf))\n",
2159
    "\n",
2160
    "    if 0 :\n",
2161
    "        conf_loss = conf_loss + (true_conf    *K.binary_crossentropy(true_conf,pred_conf))\n",
2162
    "        conf_loss = conf_loss + ((1-true_conf)*K.binary_crossentropy(1.0-true_conf,1.0-pred_conf))\n",
2163
    "\n",
2164
    "    if 1 : conf_loss = conf_loss + K.binary_crossentropy(true_conf[:],pred_conf[:])\n",
2165
    "        \n",
2166
    "    if 0 : conf_loss = K.sum(conf_loss,axis=(1,2,3)) * K.sum(K.abs(xxx1_conf))/(nobjects1+1.0)\n",
2167
    "\n",
2168
    "    #\n",
2169
    "    #  xy - zone\n",
2170
    "    #\n",
2171
    "    \n",
2172
    "    true_xy   = y_true_[...,1:3]*(shape_net[1],shape_net[0])\n",
2173
    "    pred_xy   = y_pred_[...,1:3]*true_conf2\n",
2174
    "    if sigmoidOK : pred_xy   = K.sigmoid(pred_xy)\n",
2175
    "    cons1     = K.constant(constant1)\n",
2176
    "\n",
2177
    "    if 0  :\n",
2178
    "        true_xy   = true_xy/(grid_net[1],grid_net[0])-cons1[:nbatch]\n",
2179
    "\n",
2180
    "        xy_loss   = (true_conf2*K.binary_crossentropy(true_xy,pred_xy)*\n",
2181
    "                     0.5*(2.0-y_true_[...,3:4]*y_true_[...,4:5]))\n",
2182
    "        #xy_loss   = (true_conf2*K.binary_crossentropy(true_xy,pred_xy))\n",
2183
    "        \n",
2184
    "    else :\n",
2185
    "        pred_xy   = (pred_xy+cons1[:nbatch])*(grid_net[1],grid_net[0])\n",
2186
    "        xy_loss   = (true_conf*K.sqrt(K.sum(K.square(true_xy-pred_xy),axis=-1)))\n",
2187
    "        \n",
2188
    "    \n",
2189
    "    #\n",
2190
    "    #  wh - zone\n",
2191
    "    #\n",
2192
    "    \n",
2193
    "    cons2     = K.constant(constant2)\n",
2194
    "    zero_wh   = K.cast(K.equal(y_true_[...,3:5],0.0),dtype=tf.float32) # for break log(0)\n",
2195
    "    true_wh   = y_true_[...,3:5]*cons2[:nbatch] #(shape_net[1],shape_net[0])/cons2[:nbatch]\n",
2196
    "    true_wh   = K.log(zero_wh+true_wh)\n",
2197
    "    pred_wh   = y_pred_[...,3:5]*true_conf2 # ????????????????????????????????????? 2019-08-09\n",
2198
    "\n",
2199
    "    wh_loss   = (true_conf2*K.square(true_wh-pred_wh)*0.5*(2.0-y_true_[...,3:4]*y_true_[...,4:5]))\n",
2200
    "\n",
2201
    "    #\n",
2202
    "    #  class - zone\n",
2203
    "    #\n",
2204
    "    \n",
2205
    "    if shape_out[-1]>5 :\n",
2206
    "        pred_class = y_pred_[...,5:]\n",
2207
    "        \n",
2208
    "        if 0 : \n",
2209
    "            pred_class= K.sigmoid(y_pred_[...,5:])\n",
2210
    "            if 1 : class_loss= (true_conf1*K.square(true_class-pred_class))\n",
2211
    "            if 0 : class_loss= (true_conf*K.sqrt(K.sum(K.square(true_class-pred_class),axis=-1)))\n",
2212
    "            if 0 : class_loss= (true_conf1*K.binary_crossentropy(true_class,pred_class))\n",
2213
    "        if 0 : class_loss= (true_conf1*K.binary_crossentropy(true_class,pred_class,from_logits=True))\n",
2214
    "        if 0 :\n",
2215
    "            pred_class= K.softmax(y_pred_[...,5:],axis=-1)\n",
2216
    "            class_loss= (true_conf*K.categorical_crossentropy(true_class,pred_class))\n",
2217
    "        if 0 : class_loss= (true_conf*K.categorical_crossentropy(true_class,pred_class))\n",
2218
    "        if 1 : class_loss= (true_conf*K.categorical_crossentropy(true_class,pred_class,from_logits=True))\n",
2219
    "        ###class_loss= (true_conf1*K.sparse_categorical_crossentropy(true_class,pred_class,from_logits=True))\n",
2220
    "\n",
2221
    "    #\n",
2222
    "    #  loss - zone\n",
2223
    "    #\n",
2224
    "    \n",
2225
    "    loss    = 0.0\n",
2226
    "    \n",
2227
    "    loss      = loss + K.sum(conf_loss)      * (shape_out[-2]+2.0)\n",
2228
    "    loss      = loss + K.sum(xy_loss)        #* (shape_out[-2]/2.0+1.0)\n",
2229
    "    loss      = loss + K.sum(wh_loss)\n",
2230
    "    \n",
2231
    "    if shape_out[-1]>5 : \n",
2232
    "        loss      = loss + K.sum(class_loss) #* (shape_out[-2]/3.0+1.0)\n",
2233
    "        \n",
2234
    "    if 0 : loss      = loss/(K.sum(nobjects1)+1.0) #(K.cast(nbatch,K.floatx())*nobjects)\n",
2235
    "    if 1 : loss      = loss/(K.cast(nbatch,K.floatx()))\n",
2236
    "    \n",
2237
    "    #print('ending')\n",
2238
    "    \n",
2239
    "    if debug : return(loss, conf_loss, xy_loss, wh_loss, \n",
2240
    "                      true_wh, pred_wh, \n",
2241
    "                      nbatch, \n",
2242
    "                      true_xy, pred_xy, \n",
2243
    "                      true_conf,\n",
2244
    "                      class_loss)\n",
2245
    "        \n",
2246
    "        \n",
2247
    "    return loss"
2248
   ]
2249
  },
2250
  {
2251
   "cell_type": "code",
2252
   "execution_count": null,
2253
   "metadata": {},
2254
   "outputs": [],
2255
   "source": []
2256
  },
2257
  {
2258
   "cell_type": "code",
2259
   "execution_count": 53,
2260
   "metadata": {
2261
    "scrolled": false
2262
   },
2263
   "outputs": [],
2264
   "source": [
2265
    "if 0 :\n",
2266
    "    if 1 :\n",
2267
    "        if 1 : ge  = gtrain(11,batch_size);\n",
2268
    "        xtr, ytr = ge.__next__()\n",
2269
    "    else :\n",
2270
    "        PList = [0,1,2,3,4,5]\n",
2271
    "        PList = [10,11,12,13,14,15]\n",
2272
    "        print(PList)\n",
2273
    "        PList    = PList[-batch_size:]\n",
2274
    "        xtr      = xtrain[PList]\n",
2275
    "        ytr      = yTrainCreate3(np.array(trainYYL)[PList],\n",
2276
    "                       classes, shape_net, shape_out, anchors_net, \n",
2277
    "                       [1.0,]*len(PList), \n",
2278
    "                       [np.array(shape_net).tolist(),]*len(PList))\n",
2279
    "        \n",
2280
    "    print(xtr.shape,ytr.shape)\n",
2281
    "    \n",
2282
    "    ptr      = model10.predict(xtr,batch_size=batch_size)\n",
2283
    "\n",
2284
    "    xtensor = K.variable(ytr)\n",
2285
    "    ytensor = K.variable(ptr)\n",
2286
    "    \n",
2287
    "    loss, conf_loss, xy_loss, wh_loss, true_wh, red_wh, nbatch, true_xy, pred_xy, true_conf, class_loss = yloss(xtensor, \n",
2288
    "                                                                                          ytensor, \n",
2289
    "                                                                                          debug=True)\n",
2290
    "    print('result=> loss={:10.4} conf={:10.4} xy={:10.4} wh={:10.4} class={:10.4}'.format(\n",
2291
    "          K.eval(loss),\n",
2292
    "          K.eval(K.sum(conf_loss)),\n",
2293
    "          K.eval(K.sum(xy_loss)),\n",
2294
    "          K.eval(K.sum(wh_loss)), \n",
2295
    "          K.eval(K.sum(class_loss)), \n",
2296
    "          K.eval(nbatch)),\n",
2297
    "         )\n",
2298
    "    \n",
2299
    "    true_conf, true_xy, pred_xy = K.eval(true_conf), K.eval(true_xy), K.eval(pred_xy)\n",
2300
    "    print(true_conf.shape, true_xy.shape, pred_xy.shape)\n",
2301
    "    '''    \n",
2302
    "    for ibb,bb in enumerate(true_conf) :\n",
2303
    "        for irr,rr in enumerate(bb) :\n",
2304
    "            for icc,cc in enumerate(rr) :\n",
2305
    "                for iaa,aa in enumerate(cc) :\n",
2306
    "                    if aa>0.5 :\n",
2307
    "                        print(true_xy[ibb,irr,icc,iaa],pred_xy[ibb,irr,icc,iaa])\n",
2308
    "    \n",
2309
    "    cconf_loss = K.eval(conf_loss)\n",
2310
    "    print(cconf_loss.shape,cconf_loss.min(),cconf_loss.max(),cconf_loss.mean())\n",
2311
    "    #print(K.eval(true_wh),K.eval(pred_wh))\n",
2312
    "    '''    \n",
2313
    "    for ii in range(len(xtr)) :\n",
2314
    "\n",
2315
    "        plt.figure(figsize=(15,12))\n",
2316
    "        plt.subplot(121)\n",
2317
    "        drawTrain1(xtr[ii].squeeze(),ytr[ii], shape_net, show=False, iiNumber=ii)\n",
2318
    "        plt.subplot(122)\n",
2319
    "        drawPred1 (xtr[ii].squeeze(),ptr[ii],anchors_net,grid_net,shape_net, show=False, threshold=0.5)\n",
2320
    "\n",
2321
    "        plt.show()\n",
2322
    "\n",
2323
    "    "
2324
   ]
2325
  },
2326
  {
2327
   "cell_type": "code",
2328
   "execution_count": 54,
2329
   "metadata": {},
2330
   "outputs": [],
2331
   "source": [
2332
    "if 0 :\n",
2333
    "    import importlib\n",
2334
    "    import xyCreate\n",
2335
    "    import formSubs\n",
2336
    "    import f1Subs\n",
2337
    "    importlib.reload(xyCreate)\n",
2338
    "    importlib.reload(formSubs)\n",
2339
    "    importlib.reload(f1Subs)\n",
2340
    "    from xyCreate import drawTrain1, drawPred1, yTrainCreate2\n",
2341
    "    from formSubs import formSub5\n",
2342
    "    from f1Subs import f1"
2343
   ]
2344
  },
2345
  {
2346
   "cell_type": "code",
2347
   "execution_count": null,
2348
   "metadata": {
2349
    "scrolled": false
2350
   },
2351
   "outputs": [],
2352
   "source": []
2353
  },
2354
  {
2355
   "cell_type": "code",
2356
   "execution_count": null,
2357
   "metadata": {},
2358
   "outputs": [],
2359
   "source": []
2360
  },
2361
  {
2362
   "cell_type": "code",
2363
   "execution_count": 55,
2364
   "metadata": {},
2365
   "outputs": [],
2366
   "source": [
2367
    "from albumentations import (\n",
2368
    "    HorizontalFlip, VerticalFlip,\n",
2369
    "    IAAPerspective, ShiftScaleRotate, CLAHE, RandomRotate90, RandomScale,\n",
2370
    "    Transpose, ShiftScaleRotate, Blur, OpticalDistortion, GridDistortion, HueSaturationValue,\n",
2371
    "    IAAAdditiveGaussianNoise, GaussNoise, MotionBlur, MedianBlur, IAAPiecewiseAffine,\n",
2372
    "    IAASharpen, IAAEmboss, RandomContrast, RandomBrightness, Flip, OneOf, Compose\n",
2373
    ")\n",
2374
    "\n",
2375
    "def boba_aug01 (p=.2) :\n",
2376
    "    return OneOf([RandomBrightness(), Blur()],p=p)\n",
2377
    "\n",
2378
    "def boba_aug02 (p=.2) :\n",
2379
    "    return Compose([RandomBrightness(), Blur()],p=p)\n",
2380
    "\n",
2381
    "def boba_aug03 (p=.3) :\n",
2382
    "    return OneOf([RandomBrightness()],p=p)\n",
2383
    "\n",
2384
    "def boba_aug04 (p=.3) :\n",
2385
    "    return OneOf([Blur(),RandomBrightness(),GaussNoise(),IAAAdditiveGaussianNoise()],p=p)\n",
2386
    "\n",
2387
    "boba_aug = boba_aug04()\n",
2388
    "\n",
2389
    "\n",
2390
    "from xyCreate import yTrainCreate2\n",
2391
    "def gtrain(lenData, batch_size, sample_weight=None, albu=True, shuffleOK=True, otsu=False) :\n",
2392
    "    trainYYA = trainYYL\n",
2393
    "    if (type(trainYYL)!=(type(np.zeros((1,))))) : trainYYA = np.array(trainYYL)\n",
2394
    "    shuffle      = np.arange(lenData).astype(np.int16)\n",
2395
    "    shape_in_all = [shape_net,]*len(trainYYL)\n",
2396
    "    while True :\n",
2397
    "        if shuffleOK : np.random.shuffle(shuffle)\n",
2398
    "        for ii in range(0,lenData,batch_size) :\n",
2399
    "            lshuffle   = shuffle[ii:ii+batch_size]\n",
2400
    "            xxtrain    = xtrain[lshuffle]\n",
2401
    "            yytrain    = yTrainCreate3(trainYYA[lshuffle], \n",
2402
    "                                       classes, shape_net, shape_out, anchors_net, \n",
2403
    "                                       [1.0,]*len(lshuffle), \n",
2404
    "                                       [np.array(shape_net).tolist(),]*len(lshuffle))\n",
2405
    "            if albu :\n",
2406
    "                for jj in range(len(xxtrain)) :\n",
2407
    "                    xx = xxtrain[jj].squeeze().copy()\n",
2408
    "                    xx = boba_aug(image=xx)['image']\n",
2409
    "                    xxtrain[jj] = xx.reshape(xxtrain[0].shape)\n",
2410
    "            \n",
2411
    "            if otsu :\n",
2412
    "                for jj in range(len(xxtrain)) :\n",
2413
    "                    ##xx = toOTSU(xxtrain[jj].squeeze())\n",
2414
    "                    xx = CLAHEAGray(xxtrain[jj].squeeze())\n",
2415
    "                    xxtrain[jj] = xx.reshape(xxtrain[0].shape)\n",
2416
    "            \n",
2417
    "            if sample_weight is not None :\n",
2418
    "                    wwtrain = sample_weight[lshuffle]\n",
2419
    "                    yield (xxtrain,yytrain,wwtrain)\n",
2420
    "            else :  yield (xxtrain,yytrain)\n",
2421
    "    return"
2422
   ]
2423
  },
2424
  {
2425
   "cell_type": "code",
2426
   "execution_count": null,
2427
   "metadata": {},
2428
   "outputs": [],
2429
   "source": []
2430
  },
2431
  {
2432
   "cell_type": "code",
2433
   "execution_count": null,
2434
   "metadata": {},
2435
   "outputs": [],
2436
   "source": []
2437
  },
2438
  {
2439
   "cell_type": "code",
2440
   "execution_count": 56,
2441
   "metadata": {
2442
    "scrolled": false
2443
   },
2444
   "outputs": [],
2445
   "source": [
2446
    "if 0 :\n",
2447
    "    ge       = gtrain(len(trainYYL),5, otsu=True) #batch_size);\n",
2448
    "    xtr, ytr = ge.__next__()\n",
2449
    "    print(xtr.shape,xtr.dtype,ytr.shape)\n",
2450
    "    for ii in range(len(xtr)) :\n",
2451
    "        drawTrain1(xtr[ii].squeeze(),ytr[ii],shape_net)"
2452
   ]
2453
  },
2454
  {
2455
   "cell_type": "code",
2456
   "execution_count": null,
2457
   "metadata": {},
2458
   "outputs": [],
2459
   "source": []
2460
  },
2461
  {
2462
   "cell_type": "code",
2463
   "execution_count": 57,
2464
   "metadata": {},
2465
   "outputs": [],
2466
   "source": [
2467
    "if 0 :\n",
2468
    "    tf.reset_default_graph()\n",
2469
    "    K.clear_session()"
2470
   ]
2471
  },
2472
  {
2473
   "cell_type": "code",
2474
   "execution_count": 177,
2475
   "metadata": {},
2476
   "outputs": [],
2477
   "source": [
2478
    "if 1 :\n",
2479
    "    from keras.models import load_model\n",
2480
    "\n",
2481
    "    def modelLoad (filemodel, customs={}, **kwarg) :\n",
2482
    "        model10 = load_model(filemodel,custom_objects=customs,**kwarg)\n",
2483
    "        return(model10)\n",
2484
    "    #model10 = modelLoad('../Temp/best_0000x0500xBB.hdf5',customs={'yloss':yloss})\n",
2485
    "    #model10 = modelLoad(os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1536x1024x1xT1.hdf5'),\n",
2486
    "    #                    customs={'yloss':yloss})\n",
2487
    "    \n",
2488
    "    try : del model010\n",
2489
    "    except : pass\n",
2490
    "    \n",
2491
    "    K.clear_session()\n",
2492
    "    mFile   = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')\n",
2493
    "    mFile   = os.path.join(dirModels,'best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5')\n",
2494
    "    model10 = modelLoad(mFile, customs={'yloss':yloss}, compile=True)"
2495
   ]
2496
  },
2497
  {
2498
   "cell_type": "code",
2499
   "execution_count": null,
2500
   "metadata": {},
2501
   "outputs": [],
2502
   "source": []
2503
  },
2504
  {
2505
   "cell_type": "code",
2506
   "execution_count": 178,
2507
   "metadata": {},
2508
   "outputs": [],
2509
   "source": [
2510
    "classes=classes_all\n",
2511
    "classes=[]"
2512
   ]
2513
  },
2514
  {
2515
   "cell_type": "code",
2516
   "execution_count": 226,
2517
   "metadata": {
2518
    "scrolled": false
2519
   },
2520
   "outputs": [
2521
    {
2522
     "name": "stdout",
2523
     "output_type": "stream",
2524
     "text": [
2525
      "2019-09-14 15:48:54.049832 batch_size= 4 0\n",
2526
      "optimizer lr = 1e-05\n",
2527
      "Epoch 161/175\n",
2528
      " - 1289s - loss: 158.6158\n",
2529
      "\n",
2530
      "Epoch 00161: loss improved from inf to 158.61669, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2531
      "Epoch 162/175\n",
2532
      " - 1285s - loss: 155.9326\n",
2533
      "\n",
2534
      "Epoch 00162: loss improved from 158.61669 to 155.92861, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2535
      "Epoch 163/175\n",
2536
      " - 1285s - loss: 156.3431\n",
2537
      "\n",
2538
      "Epoch 00163: loss did not improve from 155.92861\n",
2539
      "Epoch 164/175\n",
2540
      " - 1285s - loss: 155.4164\n",
2541
      "\n",
2542
      "Epoch 00164: loss improved from 155.92861 to 155.37560, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2543
      "Epoch 165/175\n",
2544
      " - 1285s - loss: 154.1784\n",
2545
      "\n",
2546
      "Epoch 00165: loss improved from 155.37560 to 154.18536, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2547
      "Epoch 166/175\n",
2548
      " - 1285s - loss: 156.0131\n",
2549
      "\n",
2550
      "Epoch 00166: loss did not improve from 154.18536\n",
2551
      "Epoch 167/175\n",
2552
      " - 1285s - loss: 154.1652\n",
2553
      "\n",
2554
      "Epoch 00167: loss improved from 154.18536 to 154.15985, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2555
      "Epoch 168/175\n",
2556
      " - 1285s - loss: 153.7296\n",
2557
      "\n",
2558
      "Epoch 00168: loss improved from 154.15985 to 153.73935, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2559
      "Epoch 169/175\n",
2560
      " - 1285s - loss: 152.5938\n",
2561
      "\n",
2562
      "Epoch 00169: loss improved from 153.73935 to 152.59351, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2563
      "Epoch 170/175\n",
2564
      " - 1285s - loss: 153.4352\n",
2565
      "\n",
2566
      "Epoch 00170: loss did not improve from 152.59351\n",
2567
      "Epoch 171/175\n",
2568
      " - 1285s - loss: 152.2745\n",
2569
      "\n",
2570
      "Epoch 00171: loss improved from 152.59351 to 152.28606, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2571
      "Epoch 172/175\n",
2572
      " - 1285s - loss: 151.9311\n",
2573
      "\n",
2574
      "Epoch 00172: loss improved from 152.28606 to 151.91148, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2575
      "Epoch 173/175\n",
2576
      " - 1285s - loss: 151.5254\n",
2577
      "\n",
2578
      "Epoch 00173: loss improved from 151.91148 to 151.51781, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2579
      "Epoch 174/175\n",
2580
      " - 1285s - loss: 151.5104\n",
2581
      "\n",
2582
      "Epoch 00174: loss improved from 151.51781 to 151.51394, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2583
      "Epoch 175/175\n",
2584
      " - 1285s - loss: 151.0157\n",
2585
      "\n",
2586
      "Epoch 00175: loss improved from 151.51394 to 151.00477, saving model to ../Model/best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5\n",
2587
      "2019-09-14 21:10:32.012811\n"
2588
     ]
2589
    }
2590
   ],
2591
   "source": [
2592
    "if 1 :\n",
2593
    "    if 0 : model10 = myYoloV3Tiny(xtrain[0].shape, num_classes=len(classes), anchors=anchors)\n",
2594
    "    model10.compile('Adam',loss=yloss)\n",
2595
    "    #model10.compile('SGD',loss=yloss)\n",
2596
    "    \n",
2597
    "print(datetime.datetime.now(),'batch_size=',batch_size,len(classes))\n",
2598
    "\n",
2599
    "fnA = 'best_ZZ_{:s}x{:04d}x{shape[0]:d}x{shape[1]:d}x{shape[2]:d}xT1.hdf5'.format(model10.name,\n",
2600
    "                                                                          len(classes),shape=xtrain.shape[1:])\n",
2601
    "fnB = 'best_ZZ_{:s}x{:04d}x{shape[0]:d}x{shape[1]:d}x{shape[2]:d}xT2.hdf5'.format(model10.name,\n",
2602
    "                                                                          len(classes),shape=xtrain.shape[1:])\n",
2603
    "fnA = os.path.join(dirModels,fnA)\n",
2604
    "fnB = os.path.join(dirModels,fnB)\n",
2605
    "    \n",
2606
    "cp1       = ModelCheckpoint(fnA,monitor='loss', save_best_only=True, mode='min', verbose=1) \n",
2607
    "\n",
2608
    "#cp2       = ModelCheckpoint(fnB,monitor='val_loss', save_best_only=True, mode='min', verbose=1) \n",
2609
    "\n",
2610
    "early     = EarlyStopping(monitor='loss', patience=10)\n",
2611
    "\n",
2612
    "lr1       = ReduceLROnPlateau(monitor='loss',     factor=0.1, patience=5,   min_lr=1e-8, verbose=1)\n",
2613
    "#lr2       = ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=7,  min_lr=1e-7, verbose=1)\n",
2614
    "\n",
2615
    "trainLen       = len(trainYYL)\n",
2616
    "trainEnd       = len(trainYYL)\n",
2617
    "\n",
2618
    "#trainLen=2000\n",
2619
    "\n",
2620
    "if 1 : K.set_value(model10.optimizer.lr, 1e-3)\n",
2621
    "if 1 : K.set_value(model10.optimizer.lr, 1e-4)\n",
2622
    "if 1 : K.set_value(model10.optimizer.lr, 1e-5)\n",
2623
    "print('optimizer lr =',K.get_value(model10.optimizer.lr))    \n",
2624
    "\n",
2625
    "hist1     = model10.fit_generator(gtrain(trainLen,batch_size,otsu=False), # , sample_weight=sample_weight),\n",
2626
    "                    steps_per_epoch=int(np.ceil(trainLen/batch_size)),\n",
2627
    "                    #epochs=354,\n",
2628
    "                    epochs=160+15,\n",
2629
    "                    #initial_epoch=324,\n",
2630
    "                    initial_epoch=160, #100,\n",
2631
    "                    #validation_data=(validationX,validationY),\n",
2632
    "                    #callbacks=[lr1,lr2,cp1,cp2,early],\n",
2633
    "                    callbacks=[lr1,cp1,early],\n",
2634
    "                    #max_queue_size=75,\n",
2635
    "                    verbose=2)\n",
2636
    "\n",
2637
    "print(datetime.datetime.now())"
2638
   ]
2639
  },
2640
  {
2641
   "cell_type": "code",
2642
   "execution_count": null,
2643
   "metadata": {},
2644
   "outputs": [],
2645
   "source": []
2646
  },
2647
  {
2648
   "cell_type": "code",
2649
   "execution_count": null,
2650
   "metadata": {},
2651
   "outputs": [],
2652
   "source": []
2653
  },
2654
  {
2655
   "cell_type": "code",
2656
   "execution_count": 227,
2657
   "metadata": {},
2658
   "outputs": [
2659
    {
2660
     "name": "stdout",
2661
     "output_type": "stream",
2662
     "text": [
2663
      "1e-05\n",
2664
      "1e-05\n"
2665
     ]
2666
    }
2667
   ],
2668
   "source": [
2669
    "print(K.get_value(model10.optimizer.lr))\n",
2670
    "if 0 : K.set_value(model10.optimizer.lr, 1e-4)\n",
2671
    "print(K.get_value(model10.optimizer.lr))"
2672
   ]
2673
  },
2674
  {
2675
   "cell_type": "code",
2676
   "execution_count": 228,
2677
   "metadata": {},
2678
   "outputs": [
2679
    {
2680
     "name": "stdout",
2681
     "output_type": "stream",
2682
     "text": [
2683
      "2019-09-14 21:10:50.712663\n"
2684
     ]
2685
    }
2686
   ],
2687
   "source": [
2688
    "print(datetime.datetime.now())"
2689
   ]
2690
  },
2691
  {
2692
   "cell_type": "code",
2693
   "execution_count": null,
2694
   "metadata": {},
2695
   "outputs": [],
2696
   "source": []
2697
  },
2698
  {
2699
   "cell_type": "code",
2700
   "execution_count": 229,
2701
   "metadata": {},
2702
   "outputs": [],
2703
   "source": [
2704
    "def makeDecision1  (model010,xxtrain,\n",
2705
    "                    shape_in, classes, anchors_net, \n",
2706
    "                    step=200,\n",
2707
    "                    batch_size=2,\n",
2708
    "                    printOK=False,\n",
2709
    "                    show=False,\n",
2710
    "                    **kwarg) :\n",
2711
    "    label  =  []\n",
2712
    "    \n",
2713
    "    shape_net = np.array(xxtrain.shape[1:])\n",
2714
    "    shape_out = model010.get_layer(index=-1).output_shape[1:]\n",
2715
    "    grid_net  = shape_net[:2]/shape_out[0:2]\n",
2716
    "    \n",
2717
    "    if printOK : print(datetime.datetime.now(),'Model1')\n",
2718
    "    for ii in range(0,len(xxtrain),step) :\n",
2719
    "        if printOK : print(datetime.datetime.now(),ii)\n",
2720
    "\n",
2721
    "       #print(datetime.datetime.now(),ii)\n",
2722
    "        pptrain      = model010.predict(xxtrain[ii:ii+step],batch_size=batch_size)\n",
2723
    "        labels       = formSub5 (pptrain, [1.0,]*len(pptrain), [shape_in,]*len(pptrain), \n",
2724
    "                                 classes, grid_net, anchors_net, **kwarg)\n",
2725
    "                                 #classesOK=True, \n",
2726
    "                                 #sigmoidOK=False)\n",
2727
    "        if show :\n",
2728
    "            for iii,xxx in enumerate(xxtrain[ii:ii+step]) :\n",
2729
    "                print(iii,len(aa),labels[iii])\n",
2730
    "                plt.imshow(xxx.squeeze()); plt.show()\n",
2731
    "\n",
2732
    "        label  = label + labels\n",
2733
    "        del pptrain\n",
2734
    "        \n",
2735
    "    return(label)\n",
2736
    "\n",
2737
    "    ################################################################\n",
2738
    "        \n",
2739
    "    model100 = modelLoad(models[1])\n",
2740
    "    shape_bb = model100.input_shape[1:]\n",
2741
    "    \n",
2742
    "    flag     = 0\n",
2743
    "\n",
2744
    "    if printOK : print(datetime.datetime.now(),'Model2')\n",
2745
    "    for jj,labelFalse in enumerate(labelsFalse) :\n",
2746
    "\n",
2747
    "        if printOK and (jj%500)==0 : print(datetime.datetime.now(),jj)\n",
2748
    "\n",
2749
    "        if labelFalse=='' : \n",
2750
    "            labels.append('')\n",
2751
    "            continue\n",
2752
    "\n",
2753
    "        if files is None :\n",
2754
    "            label, freq = bbFalseToTrue (xtrain[jj], labelFalse, classes, model100, \n",
2755
    "                                         shape_bb, scale_all[jj],\n",
2756
    "                                         batch_size=batching[1],\n",
2757
    "                                         printOK=printOK,\n",
2758
    "                                         iiNumber=jj,\n",
2759
    "                                         **kwarg)\n",
2760
    "        else :\n",
2761
    "\n",
2762
    "            nFile       = os.path.join(files[0],files[1][jj]+'.jpg')\n",
2763
    "\n",
2764
    "            #print(nFile)\n",
2765
    "\n",
2766
    "            label, freq = bbFalseToTrueF(nFile, labelFalse, classes, model100, \n",
2767
    "                                         shape_bb,\n",
2768
    "                                         batch_size=batching[1],\n",
2769
    "                                         printOK=printOK,\n",
2770
    "                                         iiNumber=jj,\n",
2771
    "                                         **kwarg) #, centerXY=False)\n",
2772
    "\n",
2773
    "        labels.append(label)\n",
2774
    "\n",
2775
    "        if flag==0 : freq0 = np.zeros(freq[0].shape,dtype=int)\n",
2776
    "            \n",
2777
    "        freq0 = freq0 + freq[0]\n",
2778
    "        freq1 = np.hstack([np.zeros((len(freq[1]),1))+ii+jj,freq[1]])\n",
2779
    "        cs.append(freq1)\n",
2780
    "        flag = flag +1\n",
2781
    "\n",
2782
    "    if flag==0 : return(labels, (None, None))\n",
2783
    "        \n",
2784
    "    cs = np.vstack(cs)\n",
2785
    "    if printOK : print(datetime.datetime.now(),'End')\n",
2786
    "    \n",
2787
    "    return(labels, (freq0,cs,labelsFalse))\n",
2788
    "\n",
2789
    "def rescale (labelsD, deltas, scale) :\n",
2790
    "    labels = ''\n",
2791
    "    for label,delta in zip(labelsD,deltas) :\n",
2792
    "        if label=='' : continue\n",
2793
    "        bb  = np.array(label.split()).reshape((-1,5))\n",
2794
    "        \n",
2795
    "        bb1 = ((bb[:,1:5].astype(float)+(delta[1],delta[0],0,0))/scale).astype(int).astype(str)\n",
2796
    "        bb[:,1:] = bb1\n",
2797
    "        labels = labels + ' ' + ' '.join([' '.join(bbb) for bbb in bb.tolist()])\n",
2798
    "    labels = labels.strip()\n",
2799
    "    return(labels)\n",
2800
    "\n",
2801
    "def makeDecisionF (mFile, \n",
2802
    "                   dirFile, nameFiles, scale, \n",
2803
    "                   shape_net, classes, anchors_net, \n",
2804
    "                   contrast=False, clahe=True, show=False, **kwarg) :\n",
2805
    "    \n",
2806
    "    xdatas, xdeltas, xscale, xshape_in = xCreate2FF(dirFile, nameFiles, shape_net, \n",
2807
    "                                                    scale=scale, contrast=contrast, clahe=clahe)\n",
2808
    "    \n",
2809
    "    model010 = modelLoad(mFile,customs={'yloss':yloss},compile=False)\n",
2810
    "    \n",
2811
    "    labels   = makeDecisionA(model010,xdatas,xdeltas,xscale,xshape_in,shape_net,classes,anchors_net, **kwarg)\n",
2812
    "    \n",
2813
    "    return(labels)\n",
2814
    "\n",
2815
    "def makeDecisionA(model010, \n",
2816
    "                  xdatas, xdeltas, xscale, xshape_in, \n",
2817
    "                  shape_net, classes, anchors_net, \n",
2818
    "                  show=False, iiStats=False, debug=False,\n",
2819
    "                  **kwarg) :\n",
2820
    "    \n",
2821
    "    for ii, xdelta in enumerate(xdeltas) :\n",
2822
    "        if ii==0  : ii0, labels, labelsD = 0, [], []\n",
2823
    "        if iiStats :\n",
2824
    "            if ii%iiStats==0 : print(datetime.datetime.now(), ii)\n",
2825
    "        labelD   = makeDecision1  (model010,xdatas[ii0:ii0+len(xdelta)],\n",
2826
    "                                   xshape_in[ii], classes, anchors_net, **kwarg)\n",
2827
    "        \n",
2828
    "        if show :\n",
2829
    "            for ill, lll in enumerate(labelD) :\n",
2830
    "                drawSub1A(xdatas[ii0+ill],labelD[ill])\n",
2831
    "            \n",
2832
    "        label    = rescale(labelD,xdelta,xscale[ii])\n",
2833
    "        #print([len(ll) for ll in labelD],len(label))\n",
2834
    "\n",
2835
    "        labels.append(label)\n",
2836
    "        if debug : labelsD = labelsD + labelD\n",
2837
    "        ii0 = ii0 + len(xdelta)\n",
2838
    "    \n",
2839
    "    if debug : return(labels,labelsD)\n",
2840
    "    \n",
2841
    "    return(labels)"
2842
   ]
2843
  },
2844
  {
2845
   "cell_type": "code",
2846
   "execution_count": 230,
2847
   "metadata": {},
2848
   "outputs": [],
2849
   "source": [
2850
    "if 1 :\n",
2851
    "    import importlib\n",
2852
    "    import xyCreate\n",
2853
    "    import formSubs\n",
2854
    "    import f1Subs\n",
2855
    "    importlib.reload(xyCreate)\n",
2856
    "    importlib.reload(formSubs)\n",
2857
    "    importlib.reload(f1Subs)\n",
2858
    "    from xyCreate import drawTrain1, drawPred1, yTrainCreate2\n",
2859
    "    from formSubs import formSub5, formSub1x4\n",
2860
    "    from f1Subs   import f1, f1All"
2861
   ]
2862
  },
2863
  {
2864
   "cell_type": "code",
2865
   "execution_count": 231,
2866
   "metadata": {},
2867
   "outputs": [],
2868
   "source": [
2869
    "if 0 :\n",
2870
    "    from keras.models import load_model\n",
2871
    "\n",
2872
    "    def modelLoad (filemodel, customs={}, **kwarg) :\n",
2873
    "        model10 = load_model(filemodel,compile=False,custom_objects=customs)\n",
2874
    "        return(model10)\n",
2875
    "    #model10 = modelLoad('../Temp/best_0000x0500xBB.hdf5',customs={'yloss':yloss})\n",
2876
    "    #model10 = modelLoad(os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx0000x1536x1024x1xT1.hdf5'),\n",
2877
    "    #                    customs={'yloss':yloss})"
2878
   ]
2879
  },
2880
  {
2881
   "cell_type": "code",
2882
   "execution_count": 232,
2883
   "metadata": {},
2884
   "outputs": [
2885
    {
2886
     "data": {
2887
      "text/plain": [
2888
       "3881"
2889
      ]
2890
     },
2891
     "execution_count": 232,
2892
     "metadata": {},
2893
     "output_type": "execute_result"
2894
    }
2895
   ],
2896
   "source": [
2897
    "len(xdeltas)"
2898
   ]
2899
  },
2900
  {
2901
   "cell_type": "code",
2902
   "execution_count": 233,
2903
   "metadata": {
2904
    "scrolled": false
2905
   },
2906
   "outputs": [
2907
    {
2908
     "name": "stdout",
2909
     "output_type": "stream",
2910
     "text": [
2911
      "2019-09-14 21:11:01.881329 0\n",
2912
      "2019-09-14 21:13:04.458718 500\n",
2913
      "2019-09-14 21:15:10.771889 1000\n",
2914
      "2019-09-14 21:17:14.678761 1500\n",
2915
      "2019-09-14 21:19:19.389363 2000\n",
2916
      "2019-09-14 21:21:23.037626 2500\n",
2917
      "2019-09-14 21:23:29.649451 3000\n",
2918
      "2019-09-14 21:25:35.353534 3500\n",
2919
      "2019-09-14 21:27:14.515601 3881 3881 7694\n"
2920
     ]
2921
    }
2922
   ],
2923
   "source": [
2924
    "if 1 :\n",
2925
    "    \n",
2926
    "    if 0 :\n",
2927
    "    \n",
2928
    "        mFile = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')    \n",
2929
    "        #mFile = os.path.join(dirModels,'best_BB_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')    \n",
2930
    "        #mFile = os.path.join(dirModels,'best_CC_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')    \n",
2931
    "\n",
2932
    "        #mFile = os.path.join(dirModels,'best_ZZ_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')    \n",
2933
    "\n",
2934
    "        try : del model010\n",
2935
    "        except : pass\n",
2936
    "\n",
2937
    "        mFile = os.path.join(dirModels,'best_DD_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')    \n",
2938
    "        mFile = os.path.join(dirModels,'best_ZZ_my-YoloV3-Tinyx0000x1024x1280x1xT1.hdf5')    \n",
2939
    "\n",
2940
    "        model010 = modelLoad(mFile,customs={'yloss':yloss},compile=False)\n",
2941
    "\n",
2942
    "    if 0 :\n",
2943
    "        nn    = 1000 # len(xdeltas)\n",
2944
    "        xnn   = np.sum([len(delta) for delta in xdeltas[:nn]])\n",
2945
    "        print('imgs={} subImgs={}'.format(nn,xnn))\n",
2946
    "\n",
2947
    "        labelsTrue, labelsD = makeDecisionA (model010,\n",
2948
    "                                    xtrain[:xnn],xdeltas[:nn],scale_all[:nn],shape_in_all[:nn],\n",
2949
    "                                    shape_net,classes,anchors_net, iiStats=500, sigmoidOK=True, debug=True)\n",
2950
    "        print(datetime.datetime.now(),len(labelsTrue),nn,xnn)\n",
2951
    "    \n",
2952
    "    else :\n",
2953
    "    \n",
2954
    "        labelsTrue = makeDecisionA (model10, # model010,\n",
2955
    "                                    xtrain,xdeltas,scale_all,shape_in_all,\n",
2956
    "                                    shape_net,classes,anchors_net, iiStats=500)\n",
2957
    "    \n",
2958
    "    \n",
2959
    "    \n",
2960
    "        print(datetime.datetime.now(),len(labelsTrue),len(xdeltas),len(xtrain))\n",
2961
    "    \n",
2962
    "    if 0 : \n",
2963
    "        for ii in range(nn) :\n",
2964
    "            fName = os.path.join(dirTrain,nameFiles[ii]+'.jpg')\n",
2965
    "            print(scale_all[ii],shape_in_all[ii],xdeltas[ii])\n",
2966
    "            plt.figure(figsize=(15,12))\n",
2967
    "            plt.subplot(121)\n",
2968
    "            drawSub1F(fName,trainYY.iloc[ii].fillna('').labels, centerXY=False, show=False)\n",
2969
    "            plt.subplot(122)\n",
2970
    "            drawSub1F(fName,labelsTrue[ii], show=False)\n",
2971
    "\n",
2972
    "            plt.show()\n"
2973
   ]
2974
  },
2975
  {
2976
   "cell_type": "code",
2977
   "execution_count": 234,
2978
   "metadata": {},
2979
   "outputs": [
2980
    {
2981
     "data": {
2982
      "text/plain": [
2983
       "(7694, 3881)"
2984
      ]
2985
     },
2986
     "execution_count": 234,
2987
     "metadata": {},
2988
     "output_type": "execute_result"
2989
    }
2990
   ],
2991
   "source": [
2992
    "len(xtrain),len(shape_in_all)"
2993
   ]
2994
  },
2995
  {
2996
   "cell_type": "code",
2997
   "execution_count": null,
2998
   "metadata": {},
2999
   "outputs": [],
3000
   "source": []
3001
  },
3002
  {
3003
   "cell_type": "code",
3004
   "execution_count": null,
3005
   "metadata": {},
3006
   "outputs": [],
3007
   "source": []
3008
  },
3009
  {
3010
   "cell_type": "code",
3011
   "execution_count": 207,
3012
   "metadata": {
3013
    "scrolled": false
3014
   },
3015
   "outputs": [],
3016
   "source": [
3017
    "if 0 :\n",
3018
    "    K.clear_session()\n",
3019
    "    mFile = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1536x1024x1xT1.hdf5')\n",
3020
    "    mFile = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1536x1024x1xT1.hdf5')\n",
3021
    "\n",
3022
    "    mFile = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')\n",
3023
    "    \n",
3024
    "    nn    = 1000\n",
3025
    "    \n",
3026
    "    labelsTrue = makeDecisionF (mFile,\n",
3027
    "                                dirTrain,nameFiles[:nn],\n",
3028
    "                                scale,shape_net,classes,anchors_net,\n",
3029
    "                                contrast=False,  clahe=True, iiStats=250)\n",
3030
    "                     \n",
3031
    "        \n",
3032
    "    if 0 : \n",
3033
    "        for ii in range(nn) :\n",
3034
    "            fName = os.path.join(dirTrain,nameFiles[ii]+'.jpg')\n",
3035
    "            plt.figure(figsize=(15,12))\n",
3036
    "            plt.subplot(121)\n",
3037
    "            drawSub1F(fName,trainYY.iloc[ii].fillna('').labels, centerXY=False, show=False)\n",
3038
    "            plt.subplot(122)\n",
3039
    "            drawSub1F(fName,labelsTrue[ii], show=False)\n",
3040
    "\n",
3041
    "            plt.show()"
3042
   ]
3043
  },
3044
  {
3045
   "cell_type": "code",
3046
   "execution_count": null,
3047
   "metadata": {},
3048
   "outputs": [],
3049
   "source": []
3050
  },
3051
  {
3052
   "cell_type": "code",
3053
   "execution_count": 235,
3054
   "metadata": {},
3055
   "outputs": [
3056
    {
3057
     "name": "stdout",
3058
     "output_type": "stream",
3059
     "text": [
3060
      "0.9965179076937519 0\n"
3061
     ]
3062
    }
3063
   ],
3064
   "source": [
3065
    "if 0 :\n",
3066
    "    print(f1(trainYYL[:len(labelsD)],labelsD,classesOK=False),\n",
3067
    "          f1(trainYYL[:len(labelsD)],labelsD,classesOK=True));\n",
3068
    "\n",
3069
    "print(f1(trainYY.fillna('').labels[:len(labelsTrue)],labelsTrue,classesOK=False),\n",
3070
    "      f1(trainYY.fillna('').labels[:len(labelsTrue)],labelsTrue,classesOK=True));"
3071
   ]
3072
  },
3073
  {
3074
   "cell_type": "code",
3075
   "execution_count": null,
3076
   "metadata": {},
3077
   "outputs": [],
3078
   "source": []
3079
  },
3080
  {
3081
   "cell_type": "code",
3082
   "execution_count": null,
3083
   "metadata": {},
3084
   "outputs": [],
3085
   "source": []
3086
  },
3087
  {
3088
   "cell_type": "code",
3089
   "execution_count": 237,
3090
   "metadata": {},
3091
   "outputs": [
3092
    {
3093
     "data": {
3094
      "text/plain": [
3095
       "(650, 650)"
3096
      ]
3097
     },
3098
     "execution_count": 237,
3099
     "metadata": {},
3100
     "output_type": "execute_result"
3101
    }
3102
   ],
3103
   "source": [
3104
    "len(labelsTrue[2].split()), len(trainYY.fillna('').labels.values[2].split())"
3105
   ]
3106
  },
3107
  {
3108
   "cell_type": "code",
3109
   "execution_count": 204,
3110
   "metadata": {},
3111
   "outputs": [
3112
    {
3113
     "data": {
3114
      "text/plain": [
3115
       "array([], dtype=float32)"
3116
      ]
3117
     },
3118
     "execution_count": 204,
3119
     "metadata": {},
3120
     "output_type": "execute_result"
3121
    }
3122
   ],
3123
   "source": [
3124
    "aa = np.where(constant1>=40); constant1[aa[0],aa[1],aa[2],aa[3],aa[4]]"
3125
   ]
3126
  },
3127
  {
3128
   "cell_type": "code",
3129
   "execution_count": 578,
3130
   "metadata": {},
3131
   "outputs": [
3132
    {
3133
     "data": {
3134
      "text/plain": [
3135
       "7.561245271682739"
3136
      ]
3137
     },
3138
     "execution_count": 578,
3139
     "metadata": {},
3140
     "output_type": "execute_result"
3141
    }
3142
   ],
3143
   "source": [
3144
    "aaa=model010.evaluate_generator(gtrain(1000,batch_size),steps=int(np.ceil(1000/batch_size))); aaa"
3145
   ]
3146
  },
3147
  {
3148
   "cell_type": "code",
3149
   "execution_count": null,
3150
   "metadata": {},
3151
   "outputs": [],
3152
   "source": []
3153
  },
3154
  {
3155
   "cell_type": "code",
3156
   "execution_count": null,
3157
   "metadata": {},
3158
   "outputs": [],
3159
   "source": []
3160
  },
3161
  {
3162
   "cell_type": "code",
3163
   "execution_count": null,
3164
   "metadata": {},
3165
   "outputs": [],
3166
   "source": []
3167
  },
3168
  {
3169
   "cell_type": "code",
3170
   "execution_count": null,
3171
   "metadata": {},
3172
   "outputs": [],
3173
   "source": []
3174
  },
3175
  {
3176
   "cell_type": "code",
3177
   "execution_count": null,
3178
   "metadata": {},
3179
   "outputs": [],
3180
   "source": []
3181
  },
3182
  {
3183
   "cell_type": "code",
3184
   "execution_count": null,
3185
   "metadata": {},
3186
   "outputs": [],
3187
   "source": [
3188
    "if 1 :\n",
3189
    "    K.clear_session()\n",
3190
    "    mFile = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1536x1024x1xT1.hdf5')\n",
3191
    "    mFile = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1536x1024x1xT1.hdf5')\n",
3192
    "\n",
3193
    "    mFile = os.path.join(dirModels,'best_AA_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')\n",
3194
    "    \n",
3195
    "    mFile = os.path.join(dirModels,'best_CC_my-YoloV3-Tinyx4212x1024x1280x1xT1.hdf5')\n",
3196
    "    \n",
3197
    "    if 0 :\n",
3198
    "    \n",
3199
    "        nn    = 1000\n",
3200
    "\n",
3201
    "        labelsTest = makeDecisionF (mFile,\n",
3202
    "                                    dirTrain,nameFiles[:nn],\n",
3203
    "                                    scale,shape_net,classes,anchors_net,\n",
3204
    "                                    contrast=False,  clahe=True, iiStats=500)\n",
3205
    "        \n",
3206
    "    else :\n",
3207
    "\n",
3208
    "        labelsTest = makeDecisionF (mFile,\n",
3209
    "                                    dirTrain,nameFiles,\n",
3210
    "                                    scale,shape_net,classes,anchors_net,\n",
3211
    "                                    contrast=False,  clahe=True, iiStats=500)\n",
3212
    "        \n",
3213
    "    if 0 : \n",
3214
    "        for ii in range(nn) :\n",
3215
    "            fName = os.path.join(dirTrain,nameFiles[ii]+'.jpg')\n",
3216
    "            plt.figure(figsize=(15,12))\n",
3217
    "            plt.subplot(121)\n",
3218
    "            drawSub1F(fName,trainYY.iloc[ii].fillna('').labels, centerXY=False, show=False)\n",
3219
    "            plt.subplot(122)\n",
3220
    "            drawSub1F(fName,labelsTrue[ii], show=False)\n",
3221
    "\n",
3222
    "            plt.show()"
3223
   ]
3224
  },
3225
  {
3226
   "cell_type": "code",
3227
   "execution_count": null,
3228
   "metadata": {},
3229
   "outputs": [],
3230
   "source": []
3231
  },
3232
  {
3233
   "cell_type": "code",
3234
   "execution_count": null,
3235
   "metadata": {},
3236
   "outputs": [],
3237
   "source": []
3238
  },
3239
  {
3240
   "cell_type": "code",
3241
   "execution_count": null,
3242
   "metadata": {},
3243
   "outputs": [],
3244
   "source": []
3245
  },
3246
  {
3247
   "cell_type": "code",
3248
   "execution_count": null,
3249
   "metadata": {},
3250
   "outputs": [],
3251
   "source": []
3252
  },
3253
  {
3254
   "cell_type": "code",
3255
   "execution_count": null,
3256
   "metadata": {},
3257
   "outputs": [],
3258
   "source": []
3259
  },
3260
  {
3261
   "cell_type": "code",
3262
   "execution_count": null,
3263
   "metadata": {},
3264
   "outputs": [],
3265
   "source": []
3266
  },
3267
  {
3268
   "cell_type": "code",
3269
   "execution_count": null,
3270
   "metadata": {},
3271
   "outputs": [],
3272
   "source": []
3273
  },
3274
  {
3275
   "cell_type": "code",
3276
   "execution_count": null,
3277
   "metadata": {},
3278
   "outputs": [],
3279
   "source": []
3280
  },
3281
  {
3282
   "cell_type": "code",
3283
   "execution_count": 209,
3284
   "metadata": {},
3285
   "outputs": [],
3286
   "source": [
3287
    "def score_page_distance(preds, truth, classesOK=True, lenPred=5):\n",
3288
    "    \"\"\"\n",
3289
    "    Scores a single page.\n",
3290
    "    Args:\n",
3291
    "        preds: prediction string of labels and center points.\n",
3292
    "        truth: ground truth string of labels and bounding boxes.\n",
3293
    "    Returns:\n",
3294
    "        True/false positive and false negative counts for the page\n",
3295
    "    \"\"\"\n",
3296
    "    \n",
3297
    "    dd = []\n",
3298
    "    \n",
3299
    "    tp = 0\n",
3300
    "    fp = 0\n",
3301
    "    fn = 0\n",
3302
    "\n",
3303
    "    truth_indices = {\n",
3304
    "        'label': 0,\n",
3305
    "        'X': 1,\n",
3306
    "        'Y': 2,\n",
3307
    "        'Width': 3,\n",
3308
    "        'Height': 4\n",
3309
    "    }\n",
3310
    "    preds_indices = {\n",
3311
    "        'label': 0,\n",
3312
    "        'X': 1,\n",
3313
    "        'Y': 2,\n",
3314
    "        'W': 3,\n",
3315
    "        'H': 4\n",
3316
    "    }\n",
3317
    "\n",
3318
    "    if pd.isna(truth) and pd.isna(preds):\n",
3319
    "        return []\n",
3320
    "        #return {'tp': tp, 'fp': fp, 'fn': fn}\n",
3321
    "\n",
3322
    "    if pd.isna(truth):\n",
3323
    "        return []\n",
3324
    "        #fp += len(preds.split(' ')) // len(preds_indices)\n",
3325
    "        #return {'tp': tp, 'fp': fp, 'fn': fn}\n",
3326
    "\n",
3327
    "    if pd.isna(preds):\n",
3328
    "        return []\n",
3329
    "        #fn += len(truth.split(' ')) // len(truth_indices)\n",
3330
    "        #return {'tp': tp, 'fp': fp, 'fn': fn}\n",
3331
    "\n",
3332
    "    if truth=='' and preds=='':\n",
3333
    "        return []\n",
3334
    "        #return {'tp': tp, 'fp': fp, 'fn': fn}\n",
3335
    "\n",
3336
    "    if truth=='':\n",
3337
    "        return []\n",
3338
    "        #fp += len(preds.split(' ')) // len(preds_indices)\n",
3339
    "        #return {'tp': tp, 'fp': fp, 'fn': fn}\n",
3340
    "\n",
3341
    "    if preds=='':\n",
3342
    "        return []\n",
3343
    "        #fn += len(truth.split(' ')) // len(truth_indices)\n",
3344
    "        #return {'tp': tp, 'fp': fp, 'fn': fn}\n",
3345
    "\n",
3346
    "    truth = truth.split(' ')\n",
3347
    "    if len(truth) % len(truth_indices) != 0:\n",
3348
    "        raise ValueError('Malformed solution string')\n",
3349
    "    truth_label = np.array(truth[truth_indices['label']::len(truth_indices)])\n",
3350
    "    truth_xmin = np.array(truth[truth_indices['X']::len(truth_indices)]).astype(float)\n",
3351
    "    truth_ymin = np.array(truth[truth_indices['Y']::len(truth_indices)]).astype(float)\n",
3352
    "    truth_xmax = truth_xmin + np.array(truth[truth_indices['Width']::len(truth_indices)]).astype(float)\n",
3353
    "    truth_ymax = truth_ymin + np.array(truth[truth_indices['Height']::len(truth_indices)]).astype(float)\n",
3354
    "\n",
3355
    "    #print(len(truth_label),truth_label[truth_label.argsort()])\n",
3356
    "    \n",
3357
    "    preds = preds.split(' ')\n",
3358
    "    if len(preds) % lenPred != 0:\n",
3359
    "        \n",
3360
    "        raise ValueError('Malformed prediction string')\n",
3361
    "    preds_label = np.array(preds[preds_indices['label']::lenPred])\n",
3362
    "    preds_x = np.array(preds[preds_indices['X']::lenPred]).astype(float)\n",
3363
    "    preds_y = np.array(preds[preds_indices['Y']::lenPred]).astype(float)\n",
3364
    "    \n",
3365
    "    preds_w = np.array(preds[preds_indices['W']::lenPred]).astype(float)\n",
3366
    "    preds_h = np.array(preds[preds_indices['H']::lenPred]).astype(float)\n",
3367
    "    \n",
3368
    "    preds_unused = np.ones(len(preds_label)).astype(bool)\n",
3369
    "    \n",
3370
    "    #print(len(preds_label),preds_label[preds_label.argsort()])\n",
3371
    "\n",
3372
    "\n",
3373
    "    for xmin, xmax, ymin, ymax, label in zip(truth_xmin, truth_xmax, truth_ymin, truth_ymax, truth_label):\n",
3374
    "        # Matching = point inside box & character same & prediction not already used\n",
3375
    "        if classesOK :\n",
3376
    "            matching = (xmin < preds_x) & (xmax > preds_x) & (ymin < preds_y) & (ymax > preds_y) & (preds_label == label) & preds_unused\n",
3377
    "        else :\n",
3378
    "            matching = (xmin < preds_x) & (xmax > preds_x) & (ymin < preds_y) & (ymax > preds_y) & preds_unused\n",
3379
    "        if matching.sum() == 0:\n",
3380
    "            fn += 1\n",
3381
    "        else:\n",
3382
    "            tp += 1\n",
3383
    "            preds_unused[np.argmax(matching)] = False\n",
3384
    "            ii = np.argmax(matching)\n",
3385
    "            xdd, ydd = (xmin+xmax)/2-preds_x[ii], (ymin+ymax)/2-preds_y[ii]\n",
3386
    "            dd.append([xdd,ydd,\n",
3387
    "                       preds_x[ii],preds_y[ii],\n",
3388
    "                       xmax-xmin,ymax-ymin,0,\n",
3389
    "                       0,0,\n",
3390
    "                       preds_w[ii],preds_h[ii]])\n",
3391
    "        #print(xmin,xmax,ymin,ymax,np.argmax(matching),matching.sum(),tp,fn,matching.any())\n",
3392
    "    fp += preds_unused.sum()\n",
3393
    "    return dd"
3394
   ]
3395
  },
3396
  {
3397
   "cell_type": "code",
3398
   "execution_count": null,
3399
   "metadata": {},
3400
   "outputs": [],
3401
   "source": []
3402
  },
3403
  {
3404
   "cell_type": "code",
3405
   "execution_count": 236,
3406
   "metadata": {
3407
    "scrolled": false
3408
   },
3409
   "outputs": [],
3410
   "source": [
3411
    "dd = []\n",
3412
    "for ii in range(len(labelsTrue)) :\n",
3413
    "    ddd = score_page_distance(labelsTrue[ii], trainYY.labels.fillna('').values[ii], classesOK=False);\n",
3414
    "    dd.append(ddd)"
3415
   ]
3416
  },
3417
  {
3418
   "cell_type": "code",
3419
   "execution_count": 237,
3420
   "metadata": {},
3421
   "outputs": [],
3422
   "source": [
3423
    "d = []\n",
3424
    "for iii, ddd in enumerate(dd) :\n",
3425
    "    if len(ddd)==0 : continue\n",
3426
    "    ddd = np.array(ddd); \n",
3427
    "    ddd[..., 6] =scale_all[iii]\n",
3428
    "    # delta xy\n",
3429
    "    ddd[..., 7] =ddd[..., 0]/ddd[...,4]\n",
3430
    "    ddd[..., 8] =ddd[..., 1]/ddd[...,5]\n",
3431
    "    # delta wh\n",
3432
    "    ddd[..., 9] =ddd[..., 9]/ddd[...,4]\n",
3433
    "    ddd[...,10] =ddd[...,10]/ddd[...,5]\n",
3434
    "    \n",
3435
    "    d.append(np.array(ddd))\n",
3436
    "d = np.vstack(d)\n",
3437
    "if 0 :\n",
3438
    "    plt.figure(figsize=(15,15))\n",
3439
    "    plt.scatter(d[...,0],d[...,1], c=d[...,6]);\n",
3440
    "    plt.colorbar()\n",
3441
    "    plt.show()\n",
3442
    "\n",
3443
    "    plt.figure(figsize=(15,15))\n",
3444
    "    plt.scatter(d[...,9],d[...,10], c=d[...,6]);\n",
3445
    "    plt.colorbar()\n",
3446
    "    plt.show()"
3447
   ]
3448
  },
3449
  {
3450
   "cell_type": "code",
3451
   "execution_count": 238,
3452
   "metadata": {},
3453
   "outputs": [
3454
    {
3455
     "data": {
3456
      "image/png": "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\n",
3457
      "text/plain": [
3458
       "<Figure size 1080x720 with 2 Axes>"
3459
      ]
3460
     },
3461
     "metadata": {
3462
      "needs_background": "light"
3463
     },
3464
     "output_type": "display_data"
3465
    },
3466
    {
3467
     "data": {
3468
      "image/png": "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\n",
3469
      "text/plain": [
3470
       "<Figure size 1080x720 with 2 Axes>"
3471
      ]
3472
     },
3473
     "metadata": {
3474
      "needs_background": "light"
3475
     },
3476
     "output_type": "display_data"
3477
    }
3478
   ],
3479
   "source": [
3480
    "xy,xedges,yedges = np.histogram2d(d[:,7],d[:,8],bins=50);\n",
3481
    "#print(xedges,yedges)\n",
3482
    "plt.figure(figsize=(15,10))\n",
3483
    "plt.clf()\n",
3484
    "#plt.imshow(xy.T,extent=[xedges[0],xedges[-1],yedges[0],yedges[-1]])\n",
3485
    "plt.imshow(xy.T,extent=[xedges[0],xedges[-1],yedges[-1],yedges[0]])\n",
3486
    "plt.grid()\n",
3487
    "plt.title(\"Delta Distance between True's and Pred's center for size square\")\n",
3488
    "plt.xlabel(\"x - value ({:6.4}-{:6.4})\".format(d[:,7].mean(),d[:,7].std()))\n",
3489
    "plt.ylabel(\"y - value ({:6.4}-{:6.4})\".format(d[:,8].mean(),d[:,8].std()))\n",
3490
    "plt.colorbar()\n",
3491
    "plt.show()\n",
3492
    "\n",
3493
    "xy,xedges,yedges = np.histogram2d(d[:,9],d[:,10],bins=150);\n",
3494
    "#print(len(xedges),len(yedges),xy.shape)\n",
3495
    "plt.figure(figsize=(15,10))\n",
3496
    "#plt.clf()\n",
3497
    "#plt.imshow(xy.T,extent=[xedges[0],xedges[-1],yedges[0],yedges[-1]])\n",
3498
    "plt.imshow(xy.T,extent=[xedges[0],xedges[-1],yedges[-1],yedges[0]])\n",
3499
    "plt.grid()\n",
3500
    "plt.title(\"Delta sizes between True's and Pred's sizes (wh)\")\n",
3501
    "plt.xlabel(\"w - value ({:6.4}-{:6.4})\".format(d[:, 9].mean(),d[:, 9].std()))\n",
3502
    "plt.ylabel(\"h - value ({:6.4}-{:6.4})\".format(d[:,10].mean(),d[:,10].std()))\n",
3503
    "plt.colorbar()\n",
3504
    "plt.show()\n",
3505
    "#print(xedges,yedges)\n",
3506
    "#len(np.where(d[:,10]<3)[0])"
3507
   ]
3508
  },
3509
  {
3510
   "cell_type": "code",
3511
   "execution_count": 239,
3512
   "metadata": {},
3513
   "outputs": [
3514
    {
3515
     "data": {
3516
      "image/png": "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\n",
3517
      "text/plain": [
3518
       "<Figure size 1080x288 with 2 Axes>"
3519
      ]
3520
     },
3521
     "metadata": {
3522
      "needs_background": "light"
3523
     },
3524
     "output_type": "display_data"
3525
    }
3526
   ],
3527
   "source": [
3528
    "plt.figure(figsize=(15,4))\n",
3529
    "plt.subplot(121);plt.hist(d[:, 9],bins=150); \n",
3530
    "plt.title('W (mean:{:}-std:{:})'.format(round(d[:, 9].mean(),5),round(d[:, 9].std(),5)))\n",
3531
    "plt.subplot(122);plt.hist(d[:,10],bins=150);\n",
3532
    "plt.title('H (mean:{:}-std:{:})'.format(round(d[:,10].mean(),5),round(d[:,10].std(),5)))\n",
3533
    "plt.show()"
3534
   ]
3535
  },
3536
  {
3537
   "cell_type": "code",
3538
   "execution_count": null,
3539
   "metadata": {},
3540
   "outputs": [],
3541
   "source": []
3542
  },
3543
  {
3544
   "cell_type": "code",
3545
   "execution_count": null,
3546
   "metadata": {},
3547
   "outputs": [],
3548
   "source": []
3549
  },
3550
  {
3551
   "cell_type": "code",
3552
   "execution_count": 228,
3553
   "metadata": {},
3554
   "outputs": [],
3555
   "source": [
3556
    "def optR (v,f,optStop=0.02) :\n",
3557
    "    if (v[2]-v[0])<optStop :\n",
3558
    "        return v[np.argmax(f)], f[np.argmax(f)]\n",
3559
    "    print(datetime.datetime.now(),v,np.round(f,5))\n",
3560
    "    v = [v[0],(v[1]+v[0])/2,v[1],(v[1]+v[2])/2,v[2]]\n",
3561
    "    fA= f1(trainYY.labels.fillna(''),\n",
3562
    "       [formSub1(pt, scale_all[ii], shape_in_all[ii], classes, threshold=v[1]) for ii,pt in enumerate(ptrain)], \n",
3563
    "       classesOK=True)\n",
3564
    "    fB= f1(trainYY.labels.fillna(''),\n",
3565
    "       [formSub1(pt, scale_all[ii], shape_in_all[ii], classes, threshold=v[3]) for ii,pt in enumerate(ptrain)], \n",
3566
    "       classesOK=True)\n",
3567
    "    f = [f[0],fA,f[1],fB,f[2]]\n",
3568
    "    imax = np.argmax(f)\n",
3569
    "    if   imax==0 :\n",
3570
    "         return(optR(v[:3],f[:3]))\n",
3571
    "    elif imax==4 : \n",
3572
    "         return(optR(v[2:],f[2:]))\n",
3573
    "    else : \n",
3574
    "         return(optR([v[imax-1],v[imax],v[imax+1]],[f[imax-1],f[imax],f[imax+1]]))\n",
3575
    "    \n",
3576
    "          \n",
3577
    "    \n",
3578
    "def opt  (vmin,vmax) :\n",
3579
    "    vmmm = (vmin+vmax)/2\n",
3580
    "    fmin= f1(trainYY.labels.fillna(''),\n",
3581
    "       [formSub1(pt, scale_all[ii], shape_in_all[ii], classes, threshold=vmin) for ii,pt in enumerate(ptrain)], \n",
3582
    "       classesOK=True)\n",
3583
    "    fmmm= f1(trainYY.labels.fillna(''),\n",
3584
    "       [formSub1(pt, scale_all[ii], shape_in_all[ii], classes, threshold=vmmm) for ii,pt in enumerate(ptrain)], \n",
3585
    "       classesOK=True)\n",
3586
    "    fmax= f1(trainYY.labels.fillna(''),\n",
3587
    "       [formSub1(pt, scale_all[ii], shape_in_all[ii], classes, threshold=vmax) for ii,pt in enumerate(ptrain)], \n",
3588
    "       classesOK=True)\n",
3589
    "    return optR ([vmin,vmmm,vmax],[fmin,fmmm,fmax])\n",
3590
    "    \n",
3591
    "    \n",
3592
    "#opt(0.01,0.99)"
3593
   ]
3594
  },
3595
  {
3596
   "cell_type": "code",
3597
   "execution_count": 229,
3598
   "metadata": {},
3599
   "outputs": [],
3600
   "source": [
3601
    "#batch_size=8"
3602
   ]
3603
  },
3604
  {
3605
   "cell_type": "code",
3606
   "execution_count": null,
3607
   "metadata": {},
3608
   "outputs": [],
3609
   "source": []
3610
  },
3611
  {
3612
   "cell_type": "code",
3613
   "execution_count": null,
3614
   "metadata": {},
3615
   "outputs": [],
3616
   "source": []
3617
  },
3618
  {
3619
   "cell_type": "code",
3620
   "execution_count": null,
3621
   "metadata": {},
3622
   "outputs": [],
3623
   "source": []
3624
  },
3625
  {
3626
   "cell_type": "code",
3627
   "execution_count": null,
3628
   "metadata": {},
3629
   "outputs": [],
3630
   "source": []
3631
  },
3632
  {
3633
   "cell_type": "code",
3634
   "execution_count": null,
3635
   "metadata": {},
3636
   "outputs": [],
3637
   "source": []
3638
  },
3639
  {
3640
   "cell_type": "code",
3641
   "execution_count": 314,
3642
   "metadata": {},
3643
   "outputs": [
3644
    {
3645
     "data": {
3646
      "text/plain": [
3647
       "0.8855429479525428"
3648
      ]
3649
     },
3650
     "execution_count": 314,
3651
     "metadata": {},
3652
     "output_type": "execute_result"
3653
    }
3654
   ],
3655
   "source": [
3656
    "f1(trainYY.labels.fillna(''),labelsFalse,classesOK=False)"
3657
   ]
3658
  },
3659
  {
3660
   "cell_type": "code",
3661
   "execution_count": 139,
3662
   "metadata": {},
3663
   "outputs": [
3664
    {
3665
     "data": {
3666
      "text/plain": [
3667
       "0.0005067753663104551"
3668
      ]
3669
     },
3670
     "execution_count": 139,
3671
     "metadata": {},
3672
     "output_type": "execute_result"
3673
    }
3674
   ],
3675
   "source": [
3676
    "f1(trainYY.labels.fillna(''), labelsTrue, classesOK=True)"
3677
   ]
3678
  },
3679
  {
3680
   "cell_type": "code",
3681
   "execution_count": 317,
3682
   "metadata": {},
3683
   "outputs": [],
3684
   "source": [
3685
    "f1train=f1All(trainYY.labels.fillna(''), labelsFalse,classesOK=False)\n",
3686
    "f1train=np.array(f1train)"
3687
   ]
3688
  },
3689
  {
3690
   "cell_type": "code",
3691
   "execution_count": null,
3692
   "metadata": {},
3693
   "outputs": [],
3694
   "source": []
3695
  },
3696
  {
3697
   "cell_type": "code",
3698
   "execution_count": 318,
3699
   "metadata": {
3700
    "scrolled": false
3701
   },
3702
   "outputs": [
3703
    {
3704
     "data": {
3705
      "image/png": "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\n",
3706
      "text/plain": [
3707
       "<Figure size 1080x360 with 1 Axes>"
3708
      ]
3709
     },
3710
     "metadata": {
3711
      "needs_background": "light"
3712
     },
3713
     "output_type": "display_data"
3714
    },
3715
    {
3716
     "data": {
3717
      "image/png": "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\n",
3718
      "text/plain": [
3719
       "<Figure size 1080x360 with 1 Axes>"
3720
      ]
3721
     },
3722
     "metadata": {
3723
      "needs_background": "light"
3724
     },
3725
     "output_type": "display_data"
3726
    }
3727
   ],
3728
   "source": [
3729
    "f1train=np.array(f1train)\n",
3730
    "plt.figure(figsize=(15,5))\n",
3731
    "plt.plot(f1train[f1train.argsort()]); plt.show()\n",
3732
    "plt.figure(figsize=(15,5))\n",
3733
    "plt.plot(f1train); plt.show()\n",
3734
    "plt.close()"
3735
   ]
3736
  },
3737
  {
3738
   "cell_type": "code",
3739
   "execution_count": null,
3740
   "metadata": {},
3741
   "outputs": [],
3742
   "source": []
3743
  },
3744
  {
3745
   "cell_type": "code",
3746
   "execution_count": null,
3747
   "metadata": {},
3748
   "outputs": [],
3749
   "source": []
3750
  },
3751
  {
3752
   "cell_type": "code",
3753
   "execution_count": null,
3754
   "metadata": {},
3755
   "outputs": [],
3756
   "source": []
3757
  },
3758
  {
3759
   "cell_type": "code",
3760
   "execution_count": 76,
3761
   "metadata": {},
3762
   "outputs": [
3763
    {
3764
     "data": {
3765
      "image/png": "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\n",
3766
      "text/plain": [
3767
       "<Figure size 432x288 with 1 Axes>"
3768
      ]
3769
     },
3770
     "metadata": {
3771
      "needs_background": "light"
3772
     },
3773
     "output_type": "display_data"
3774
    }
3775
   ],
3776
   "source": [
3777
    "splitA = np.digitize((f1train*10).astype(np.int32), bins=np.arange(30))\n",
3778
    "plt.hist(splitA,bins=100); plt.show()\n",
3779
    "X = np.arange(len(trainY))\n",
3780
    "#print(split.shape, split[:100], X[:100])\n",
3781
    "from sklearn.model_selection import train_test_split\n",
3782
    "x_train, x_test = train_test_split(X, test_size=0.15, stratify=splitA) #, random_state=311)\n",
3783
    "splitA = np.hstack([x_train,x_test])\n",
3784
    "#np.save('../Work/split.npy',split)"
3785
   ]
3786
  },
3787
  {
3788
   "cell_type": "code",
3789
   "execution_count": null,
3790
   "metadata": {},
3791
   "outputs": [],
3792
   "source": []
3793
  },
3794
  {
3795
   "cell_type": "code",
3796
   "execution_count": 90,
3797
   "metadata": {},
3798
   "outputs": [
3799
    {
3800
     "data": {
3801
      "text/plain": [
3802
       "(array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15, 16,\n",
3803
       "        17, 18, 19], dtype=int32),\n",
3804
       " array([3383,  487,  297, 2388,  304,  401,   31, 2576, 1576, 1583, 2139,\n",
3805
       "        1597, 2317, 3407,  845, 1940, 3237, 2100, 1607,  942]),\n",
3806
       " array([3383,  487,  297, 2388,  304,  401,   31, 2576, 1576, 1583, 2139,\n",
3807
       "        1597, 2317, 3407,  845, 1940, 3237, 2100, 1607,  942], dtype=int32))"
3808
      ]
3809
     },
3810
     "execution_count": 90,
3811
     "metadata": {},
3812
     "output_type": "execute_result"
3813
    }
3814
   ],
3815
   "source": [
3816
    "split[:20], splitA[:20], split[splitA][:20]"
3817
   ]
3818
  },
3819
  {
3820
   "cell_type": "code",
3821
   "execution_count": 91,
3822
   "metadata": {},
3823
   "outputs": [],
3824
   "source": [
3825
    "split = split[splitA]"
3826
   ]
3827
  },
3828
  {
3829
   "cell_type": "code",
3830
   "execution_count": null,
3831
   "metadata": {},
3832
   "outputs": [],
3833
   "source": []
3834
  },
3835
  {
3836
   "cell_type": "code",
3837
   "execution_count": null,
3838
   "metadata": {},
3839
   "outputs": [],
3840
   "source": [
3841
    "#####################################################################################################"
3842
   ]
3843
  },
3844
  {
3845
   "cell_type": "code",
3846
   "execution_count": null,
3847
   "metadata": {},
3848
   "outputs": [],
3849
   "source": []
3850
  },
3851
  {
3852
   "cell_type": "code",
3853
   "execution_count": 199,
3854
   "metadata": {
3855
    "scrolled": false
3856
   },
3857
   "outputs": [
3858
    {
3859
     "name": "stdout",
3860
     "output_type": "stream",
3861
     "text": [
3862
      "__________________________________________________________________________________________________\n",
3863
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
3864
      "==================================================================================================\n",
3865
      "input (InputLayer)              (None, 1024, 768, 1) 0                                            \n",
3866
      "__________________________________________________________________________________________________\n",
3867
      "conv2d_418 (Conv2D)             (None, 1024, 768, 32 320         input[0][0]                      \n",
3868
      "__________________________________________________________________________________________________\n",
3869
      "conv2d_419 (Conv2D)             (None, 512, 384, 64) 18496       conv2d_418[0][0]                 \n",
3870
      "__________________________________________________________________________________________________\n",
3871
      "batch_normalization_284 (BatchN (None, 512, 384, 64) 256         conv2d_419[0][0]                 \n",
3872
      "__________________________________________________________________________________________________\n",
3873
      "activation_185 (Activation)     (None, 512, 384, 64) 0           batch_normalization_284[0][0]    \n",
3874
      "__________________________________________________________________________________________________\n",
3875
      "conv2d_420 (Conv2D)             (None, 512, 384, 64) 36928       activation_185[0][0]             \n",
3876
      "__________________________________________________________________________________________________\n",
3877
      "batch_normalization_285 (BatchN (None, 512, 384, 64) 256         conv2d_420[0][0]                 \n",
3878
      "__________________________________________________________________________________________________\n",
3879
      "activation_186 (Activation)     (None, 512, 384, 64) 0           batch_normalization_285[0][0]    \n",
3880
      "__________________________________________________________________________________________________\n",
3881
      "conv2d_421 (Conv2D)             (None, 512, 384, 64) 36928       activation_186[0][0]             \n",
3882
      "__________________________________________________________________________________________________\n",
3883
      "add_93 (Add)                    (None, 512, 384, 64) 0           conv2d_421[0][0]                 \n",
3884
      "                                                                 conv2d_419[0][0]                 \n",
3885
      "__________________________________________________________________________________________________\n",
3886
      "conv2d_422 (Conv2D)             (None, 256, 192, 128 73856       add_93[0][0]                     \n",
3887
      "__________________________________________________________________________________________________\n",
3888
      "batch_normalization_286 (BatchN (None, 256, 192, 128 512         conv2d_422[0][0]                 \n",
3889
      "__________________________________________________________________________________________________\n",
3890
      "activation_187 (Activation)     (None, 256, 192, 128 0           batch_normalization_286[0][0]    \n",
3891
      "__________________________________________________________________________________________________\n",
3892
      "conv2d_423 (Conv2D)             (None, 256, 192, 128 147584      activation_187[0][0]             \n",
3893
      "__________________________________________________________________________________________________\n",
3894
      "batch_normalization_287 (BatchN (None, 256, 192, 128 512         conv2d_423[0][0]                 \n",
3895
      "__________________________________________________________________________________________________\n",
3896
      "activation_188 (Activation)     (None, 256, 192, 128 0           batch_normalization_287[0][0]    \n",
3897
      "__________________________________________________________________________________________________\n",
3898
      "conv2d_424 (Conv2D)             (None, 256, 192, 128 147584      activation_188[0][0]             \n",
3899
      "__________________________________________________________________________________________________\n",
3900
      "add_94 (Add)                    (None, 256, 192, 128 0           conv2d_424[0][0]                 \n",
3901
      "                                                                 conv2d_422[0][0]                 \n",
3902
      "__________________________________________________________________________________________________\n",
3903
      "batch_normalization_288 (BatchN (None, 256, 192, 128 512         add_94[0][0]                     \n",
3904
      "__________________________________________________________________________________________________\n",
3905
      "activation_189 (Activation)     (None, 256, 192, 128 0           batch_normalization_288[0][0]    \n",
3906
      "__________________________________________________________________________________________________\n",
3907
      "conv2d_425 (Conv2D)             (None, 256, 192, 128 147584      activation_189[0][0]             \n",
3908
      "__________________________________________________________________________________________________\n",
3909
      "batch_normalization_289 (BatchN (None, 256, 192, 128 512         conv2d_425[0][0]                 \n",
3910
      "__________________________________________________________________________________________________\n",
3911
      "activation_190 (Activation)     (None, 256, 192, 128 0           batch_normalization_289[0][0]    \n",
3912
      "__________________________________________________________________________________________________\n",
3913
      "conv2d_426 (Conv2D)             (None, 256, 192, 128 147584      activation_190[0][0]             \n",
3914
      "__________________________________________________________________________________________________\n",
3915
      "add_95 (Add)                    (None, 256, 192, 128 0           conv2d_426[0][0]                 \n",
3916
      "                                                                 add_94[0][0]                     \n",
3917
      "__________________________________________________________________________________________________\n",
3918
      "conv2d_427 (Conv2D)             (None, 128, 96, 256) 295168      add_95[0][0]                     \n",
3919
      "__________________________________________________________________________________________________\n",
3920
      "batch_normalization_290 (BatchN (None, 128, 96, 256) 1024        conv2d_427[0][0]                 \n",
3921
      "__________________________________________________________________________________________________\n",
3922
      "activation_191 (Activation)     (None, 128, 96, 256) 0           batch_normalization_290[0][0]    \n",
3923
      "__________________________________________________________________________________________________\n",
3924
      "conv2d_428 (Conv2D)             (None, 128, 96, 256) 590080      activation_191[0][0]             \n",
3925
      "__________________________________________________________________________________________________\n",
3926
      "batch_normalization_291 (BatchN (None, 128, 96, 256) 1024        conv2d_428[0][0]                 \n",
3927
      "__________________________________________________________________________________________________\n",
3928
      "activation_192 (Activation)     (None, 128, 96, 256) 0           batch_normalization_291[0][0]    \n",
3929
      "__________________________________________________________________________________________________\n",
3930
      "conv2d_429 (Conv2D)             (None, 128, 96, 256) 590080      activation_192[0][0]             \n",
3931
      "__________________________________________________________________________________________________\n",
3932
      "add_96 (Add)                    (None, 128, 96, 256) 0           conv2d_429[0][0]                 \n",
3933
      "                                                                 conv2d_427[0][0]                 \n",
3934
      "__________________________________________________________________________________________________\n",
3935
      "batch_normalization_292 (BatchN (None, 128, 96, 256) 1024        add_96[0][0]                     \n",
3936
      "__________________________________________________________________________________________________\n",
3937
      "activation_193 (Activation)     (None, 128, 96, 256) 0           batch_normalization_292[0][0]    \n",
3938
      "__________________________________________________________________________________________________\n",
3939
      "conv2d_430 (Conv2D)             (None, 128, 96, 256) 590080      activation_193[0][0]             \n",
3940
      "__________________________________________________________________________________________________\n",
3941
      "batch_normalization_293 (BatchN (None, 128, 96, 256) 1024        conv2d_430[0][0]                 \n",
3942
      "__________________________________________________________________________________________________\n",
3943
      "activation_194 (Activation)     (None, 128, 96, 256) 0           batch_normalization_293[0][0]    \n",
3944
      "__________________________________________________________________________________________________\n",
3945
      "conv2d_431 (Conv2D)             (None, 128, 96, 256) 590080      activation_194[0][0]             \n",
3946
      "__________________________________________________________________________________________________\n",
3947
      "add_97 (Add)                    (None, 128, 96, 256) 0           conv2d_431[0][0]                 \n",
3948
      "                                                                 add_96[0][0]                     \n",
3949
      "__________________________________________________________________________________________________\n",
3950
      "batch_normalization_294 (BatchN (None, 128, 96, 256) 1024        add_97[0][0]                     \n",
3951
      "__________________________________________________________________________________________________\n",
3952
      "activation_195 (Activation)     (None, 128, 96, 256) 0           batch_normalization_294[0][0]    \n",
3953
      "__________________________________________________________________________________________________\n",
3954
      "conv2d_432 (Conv2D)             (None, 128, 96, 256) 590080      activation_195[0][0]             \n",
3955
      "__________________________________________________________________________________________________\n",
3956
      "batch_normalization_295 (BatchN (None, 128, 96, 256) 1024        conv2d_432[0][0]                 \n",
3957
      "__________________________________________________________________________________________________\n",
3958
      "activation_196 (Activation)     (None, 128, 96, 256) 0           batch_normalization_295[0][0]    \n",
3959
      "__________________________________________________________________________________________________\n",
3960
      "conv2d_433 (Conv2D)             (None, 128, 96, 256) 590080      activation_196[0][0]             \n",
3961
      "__________________________________________________________________________________________________\n",
3962
      "add_98 (Add)                    (None, 128, 96, 256) 0           conv2d_433[0][0]                 \n",
3963
      "                                                                 add_97[0][0]                     \n",
3964
      "__________________________________________________________________________________________________\n",
3965
      "batch_normalization_296 (BatchN (None, 128, 96, 256) 1024        add_98[0][0]                     \n",
3966
      "__________________________________________________________________________________________________\n",
3967
      "activation_197 (Activation)     (None, 128, 96, 256) 0           batch_normalization_296[0][0]    \n",
3968
      "__________________________________________________________________________________________________\n",
3969
      "conv2d_434 (Conv2D)             (None, 128, 96, 256) 590080      activation_197[0][0]             \n",
3970
      "__________________________________________________________________________________________________\n",
3971
      "batch_normalization_297 (BatchN (None, 128, 96, 256) 1024        conv2d_434[0][0]                 \n",
3972
      "__________________________________________________________________________________________________\n",
3973
      "activation_198 (Activation)     (None, 128, 96, 256) 0           batch_normalization_297[0][0]    \n",
3974
      "__________________________________________________________________________________________________\n",
3975
      "conv2d_435 (Conv2D)             (None, 128, 96, 256) 590080      activation_198[0][0]             \n",
3976
      "__________________________________________________________________________________________________\n",
3977
      "add_99 (Add)                    (None, 128, 96, 256) 0           conv2d_435[0][0]                 \n",
3978
      "                                                                 add_98[0][0]                     \n",
3979
      "__________________________________________________________________________________________________\n",
3980
      "batch_normalization_298 (BatchN (None, 128, 96, 256) 1024        add_99[0][0]                     \n",
3981
      "__________________________________________________________________________________________________\n",
3982
      "activation_199 (Activation)     (None, 128, 96, 256) 0           batch_normalization_298[0][0]    \n",
3983
      "__________________________________________________________________________________________________\n",
3984
      "conv2d_436 (Conv2D)             (None, 128, 96, 256) 590080      activation_199[0][0]             \n",
3985
      "__________________________________________________________________________________________________\n",
3986
      "batch_normalization_299 (BatchN (None, 128, 96, 256) 1024        conv2d_436[0][0]                 \n",
3987
      "__________________________________________________________________________________________________\n",
3988
      "activation_200 (Activation)     (None, 128, 96, 256) 0           batch_normalization_299[0][0]    \n",
3989
      "__________________________________________________________________________________________________\n",
3990
      "conv2d_437 (Conv2D)             (None, 128, 96, 256) 590080      activation_200[0][0]             \n",
3991
      "__________________________________________________________________________________________________\n",
3992
      "add_100 (Add)                   (None, 128, 96, 256) 0           conv2d_437[0][0]                 \n",
3993
      "                                                                 add_99[0][0]                     \n",
3994
      "__________________________________________________________________________________________________\n",
3995
      "batch_normalization_300 (BatchN (None, 128, 96, 256) 1024        add_100[0][0]                    \n",
3996
      "__________________________________________________________________________________________________\n",
3997
      "activation_201 (Activation)     (None, 128, 96, 256) 0           batch_normalization_300[0][0]    \n",
3998
      "__________________________________________________________________________________________________\n",
3999
      "conv2d_438 (Conv2D)             (None, 128, 96, 256) 590080      activation_201[0][0]             \n",
4000
      "__________________________________________________________________________________________________\n",
4001
      "batch_normalization_301 (BatchN (None, 128, 96, 256) 1024        conv2d_438[0][0]                 \n",
4002
      "__________________________________________________________________________________________________\n",
4003
      "activation_202 (Activation)     (None, 128, 96, 256) 0           batch_normalization_301[0][0]    \n",
4004
      "__________________________________________________________________________________________________\n",
4005
      "conv2d_439 (Conv2D)             (None, 128, 96, 256) 590080      activation_202[0][0]             \n",
4006
      "__________________________________________________________________________________________________\n",
4007
      "add_101 (Add)                   (None, 128, 96, 256) 0           conv2d_439[0][0]                 \n",
4008
      "                                                                 add_100[0][0]                    \n",
4009
      "__________________________________________________________________________________________________\n",
4010
      "batch_normalization_302 (BatchN (None, 128, 96, 256) 1024        add_101[0][0]                    \n",
4011
      "__________________________________________________________________________________________________\n",
4012
      "activation_203 (Activation)     (None, 128, 96, 256) 0           batch_normalization_302[0][0]    \n",
4013
      "__________________________________________________________________________________________________\n",
4014
      "conv2d_440 (Conv2D)             (None, 128, 96, 256) 590080      activation_203[0][0]             \n",
4015
      "__________________________________________________________________________________________________\n",
4016
      "batch_normalization_303 (BatchN (None, 128, 96, 256) 1024        conv2d_440[0][0]                 \n",
4017
      "__________________________________________________________________________________________________\n",
4018
      "activation_204 (Activation)     (None, 128, 96, 256) 0           batch_normalization_303[0][0]    \n",
4019
      "__________________________________________________________________________________________________\n",
4020
      "conv2d_441 (Conv2D)             (None, 128, 96, 256) 590080      activation_204[0][0]             \n",
4021
      "__________________________________________________________________________________________________\n",
4022
      "add_102 (Add)                   (None, 128, 96, 256) 0           conv2d_441[0][0]                 \n",
4023
      "                                                                 add_101[0][0]                    \n",
4024
      "__________________________________________________________________________________________________\n",
4025
      "batch_normalization_304 (BatchN (None, 128, 96, 256) 1024        add_102[0][0]                    \n",
4026
      "__________________________________________________________________________________________________\n",
4027
      "activation_205 (Activation)     (None, 128, 96, 256) 0           batch_normalization_304[0][0]    \n",
4028
      "__________________________________________________________________________________________________\n",
4029
      "conv2d_442 (Conv2D)             (None, 128, 96, 256) 590080      activation_205[0][0]             \n",
4030
      "__________________________________________________________________________________________________\n",
4031
      "batch_normalization_305 (BatchN (None, 128, 96, 256) 1024        conv2d_442[0][0]                 \n",
4032
      "__________________________________________________________________________________________________\n",
4033
      "activation_206 (Activation)     (None, 128, 96, 256) 0           batch_normalization_305[0][0]    \n",
4034
      "__________________________________________________________________________________________________\n",
4035
      "conv2d_443 (Conv2D)             (None, 128, 96, 256) 590080      activation_206[0][0]             \n",
4036
      "__________________________________________________________________________________________________\n",
4037
      "add_103 (Add)                   (None, 128, 96, 256) 0           conv2d_443[0][0]                 \n",
4038
      "                                                                 add_102[0][0]                    \n",
4039
      "__________________________________________________________________________________________________\n",
4040
      "conv2d_444 (Conv2D)             (None, 64, 48, 512)  1180160     add_103[0][0]                    \n",
4041
      "__________________________________________________________________________________________________\n",
4042
      "batch_normalization_306 (BatchN (None, 64, 48, 512)  2048        conv2d_444[0][0]                 \n",
4043
      "__________________________________________________________________________________________________\n",
4044
      "activation_207 (Activation)     (None, 64, 48, 512)  0           batch_normalization_306[0][0]    \n",
4045
      "__________________________________________________________________________________________________\n",
4046
      "conv2d_445 (Conv2D)             (None, 64, 48, 512)  2359808     activation_207[0][0]             \n",
4047
      "__________________________________________________________________________________________________\n",
4048
      "batch_normalization_307 (BatchN (None, 64, 48, 512)  2048        conv2d_445[0][0]                 \n",
4049
      "__________________________________________________________________________________________________\n",
4050
      "activation_208 (Activation)     (None, 64, 48, 512)  0           batch_normalization_307[0][0]    \n",
4051
      "__________________________________________________________________________________________________\n",
4052
      "conv2d_446 (Conv2D)             (None, 64, 48, 512)  2359808     activation_208[0][0]             \n",
4053
      "__________________________________________________________________________________________________\n",
4054
      "add_104 (Add)                   (None, 64, 48, 512)  0           conv2d_446[0][0]                 \n",
4055
      "                                                                 conv2d_444[0][0]                 \n",
4056
      "__________________________________________________________________________________________________\n",
4057
      "batch_normalization_308 (BatchN (None, 64, 48, 512)  2048        add_104[0][0]                    \n",
4058
      "__________________________________________________________________________________________________\n",
4059
      "activation_209 (Activation)     (None, 64, 48, 512)  0           batch_normalization_308[0][0]    \n",
4060
      "__________________________________________________________________________________________________\n",
4061
      "conv2d_447 (Conv2D)             (None, 64, 48, 512)  2359808     activation_209[0][0]             \n",
4062
      "__________________________________________________________________________________________________\n",
4063
      "batch_normalization_309 (BatchN (None, 64, 48, 512)  2048        conv2d_447[0][0]                 \n",
4064
      "__________________________________________________________________________________________________\n",
4065
      "activation_210 (Activation)     (None, 64, 48, 512)  0           batch_normalization_309[0][0]    \n",
4066
      "__________________________________________________________________________________________________\n",
4067
      "conv2d_448 (Conv2D)             (None, 64, 48, 512)  2359808     activation_210[0][0]             \n",
4068
      "__________________________________________________________________________________________________\n",
4069
      "add_105 (Add)                   (None, 64, 48, 512)  0           conv2d_448[0][0]                 \n",
4070
      "                                                                 add_104[0][0]                    \n",
4071
      "__________________________________________________________________________________________________\n",
4072
      "batch_normalization_310 (BatchN (None, 64, 48, 512)  2048        add_105[0][0]                    \n",
4073
      "__________________________________________________________________________________________________\n",
4074
      "activation_211 (Activation)     (None, 64, 48, 512)  0           batch_normalization_310[0][0]    \n",
4075
      "__________________________________________________________________________________________________\n",
4076
      "conv2d_449 (Conv2D)             (None, 64, 48, 512)  2359808     activation_211[0][0]             \n",
4077
      "__________________________________________________________________________________________________\n",
4078
      "batch_normalization_311 (BatchN (None, 64, 48, 512)  2048        conv2d_449[0][0]                 \n",
4079
      "__________________________________________________________________________________________________\n",
4080
      "activation_212 (Activation)     (None, 64, 48, 512)  0           batch_normalization_311[0][0]    \n",
4081
      "__________________________________________________________________________________________________\n",
4082
      "conv2d_450 (Conv2D)             (None, 64, 48, 512)  2359808     activation_212[0][0]             \n",
4083
      "__________________________________________________________________________________________________\n",
4084
      "add_106 (Add)                   (None, 64, 48, 512)  0           conv2d_450[0][0]                 \n",
4085
      "                                                                 add_105[0][0]                    \n",
4086
      "__________________________________________________________________________________________________\n",
4087
      "batch_normalization_312 (BatchN (None, 64, 48, 512)  2048        add_106[0][0]                    \n",
4088
      "__________________________________________________________________________________________________\n",
4089
      "activation_213 (Activation)     (None, 64, 48, 512)  0           batch_normalization_312[0][0]    \n",
4090
      "__________________________________________________________________________________________________\n",
4091
      "conv2d_451 (Conv2D)             (None, 64, 48, 512)  2359808     activation_213[0][0]             \n",
4092
      "__________________________________________________________________________________________________\n",
4093
      "batch_normalization_313 (BatchN (None, 64, 48, 512)  2048        conv2d_451[0][0]                 \n",
4094
      "__________________________________________________________________________________________________\n",
4095
      "activation_214 (Activation)     (None, 64, 48, 512)  0           batch_normalization_313[0][0]    \n",
4096
      "__________________________________________________________________________________________________\n",
4097
      "conv2d_452 (Conv2D)             (None, 64, 48, 512)  2359808     activation_214[0][0]             \n",
4098
      "__________________________________________________________________________________________________\n",
4099
      "add_107 (Add)                   (None, 64, 48, 512)  0           conv2d_452[0][0]                 \n",
4100
      "                                                                 add_106[0][0]                    \n",
4101
      "__________________________________________________________________________________________________\n",
4102
      "batch_normalization_314 (BatchN (None, 64, 48, 512)  2048        add_107[0][0]                    \n",
4103
      "__________________________________________________________________________________________________\n",
4104
      "activation_215 (Activation)     (None, 64, 48, 512)  0           batch_normalization_314[0][0]    \n",
4105
      "__________________________________________________________________________________________________\n",
4106
      "conv2d_453 (Conv2D)             (None, 64, 48, 512)  2359808     activation_215[0][0]             \n",
4107
      "__________________________________________________________________________________________________\n",
4108
      "batch_normalization_315 (BatchN (None, 64, 48, 512)  2048        conv2d_453[0][0]                 \n",
4109
      "__________________________________________________________________________________________________\n",
4110
      "activation_216 (Activation)     (None, 64, 48, 512)  0           batch_normalization_315[0][0]    \n",
4111
      "__________________________________________________________________________________________________\n",
4112
      "conv2d_454 (Conv2D)             (None, 64, 48, 512)  2359808     activation_216[0][0]             \n",
4113
      "__________________________________________________________________________________________________\n",
4114
      "add_108 (Add)                   (None, 64, 48, 512)  0           conv2d_454[0][0]                 \n",
4115
      "                                                                 add_107[0][0]                    \n",
4116
      "__________________________________________________________________________________________________\n",
4117
      "batch_normalization_316 (BatchN (None, 64, 48, 512)  2048        add_108[0][0]                    \n",
4118
      "__________________________________________________________________________________________________\n",
4119
      "activation_217 (Activation)     (None, 64, 48, 512)  0           batch_normalization_316[0][0]    \n",
4120
      "__________________________________________________________________________________________________\n",
4121
      "conv2d_455 (Conv2D)             (None, 64, 48, 512)  2359808     activation_217[0][0]             \n",
4122
      "__________________________________________________________________________________________________\n",
4123
      "batch_normalization_317 (BatchN (None, 64, 48, 512)  2048        conv2d_455[0][0]                 \n",
4124
      "__________________________________________________________________________________________________\n",
4125
      "activation_218 (Activation)     (None, 64, 48, 512)  0           batch_normalization_317[0][0]    \n",
4126
      "__________________________________________________________________________________________________\n",
4127
      "conv2d_456 (Conv2D)             (None, 64, 48, 512)  2359808     activation_218[0][0]             \n",
4128
      "__________________________________________________________________________________________________\n",
4129
      "add_109 (Add)                   (None, 64, 48, 512)  0           conv2d_456[0][0]                 \n",
4130
      "                                                                 add_108[0][0]                    \n",
4131
      "__________________________________________________________________________________________________\n",
4132
      "batch_normalization_318 (BatchN (None, 64, 48, 512)  2048        add_109[0][0]                    \n",
4133
      "__________________________________________________________________________________________________\n",
4134
      "activation_219 (Activation)     (None, 64, 48, 512)  0           batch_normalization_318[0][0]    \n",
4135
      "__________________________________________________________________________________________________\n",
4136
      "conv2d_457 (Conv2D)             (None, 64, 48, 512)  2359808     activation_219[0][0]             \n",
4137
      "__________________________________________________________________________________________________\n",
4138
      "batch_normalization_319 (BatchN (None, 64, 48, 512)  2048        conv2d_457[0][0]                 \n",
4139
      "__________________________________________________________________________________________________\n",
4140
      "activation_220 (Activation)     (None, 64, 48, 512)  0           batch_normalization_319[0][0]    \n",
4141
      "__________________________________________________________________________________________________\n",
4142
      "conv2d_458 (Conv2D)             (None, 64, 48, 512)  2359808     activation_220[0][0]             \n",
4143
      "__________________________________________________________________________________________________\n",
4144
      "add_110 (Add)                   (None, 64, 48, 512)  0           conv2d_458[0][0]                 \n",
4145
      "                                                                 add_109[0][0]                    \n",
4146
      "__________________________________________________________________________________________________\n",
4147
      "batch_normalization_320 (BatchN (None, 64, 48, 512)  2048        add_110[0][0]                    \n",
4148
      "__________________________________________________________________________________________________\n",
4149
      "activation_221 (Activation)     (None, 64, 48, 512)  0           batch_normalization_320[0][0]    \n",
4150
      "__________________________________________________________________________________________________\n",
4151
      "conv2d_459 (Conv2D)             (None, 64, 48, 512)  2359808     activation_221[0][0]             \n",
4152
      "__________________________________________________________________________________________________\n",
4153
      "batch_normalization_321 (BatchN (None, 64, 48, 512)  2048        conv2d_459[0][0]                 \n",
4154
      "__________________________________________________________________________________________________\n",
4155
      "activation_222 (Activation)     (None, 64, 48, 512)  0           batch_normalization_321[0][0]    \n",
4156
      "__________________________________________________________________________________________________\n",
4157
      "conv2d_460 (Conv2D)             (None, 64, 48, 512)  2359808     activation_222[0][0]             \n",
4158
      "__________________________________________________________________________________________________\n",
4159
      "add_111 (Add)                   (None, 64, 48, 512)  0           conv2d_460[0][0]                 \n",
4160
      "                                                                 add_110[0][0]                    \n",
4161
      "__________________________________________________________________________________________________\n",
4162
      "conv2d_461 (Conv2D)             (None, 32, 24, 1024) 4719616     add_111[0][0]                    \n",
4163
      "__________________________________________________________________________________________________\n",
4164
      "batch_normalization_322 (BatchN (None, 32, 24, 1024) 4096        conv2d_461[0][0]                 \n",
4165
      "__________________________________________________________________________________________________\n",
4166
      "activation_223 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_322[0][0]    \n",
4167
      "__________________________________________________________________________________________________\n",
4168
      "conv2d_462 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_223[0][0]             \n",
4169
      "__________________________________________________________________________________________________\n",
4170
      "batch_normalization_323 (BatchN (None, 32, 24, 1024) 4096        conv2d_462[0][0]                 \n",
4171
      "__________________________________________________________________________________________________\n",
4172
      "activation_224 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_323[0][0]    \n",
4173
      "__________________________________________________________________________________________________\n",
4174
      "conv2d_463 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_224[0][0]             \n",
4175
      "__________________________________________________________________________________________________\n",
4176
      "add_112 (Add)                   (None, 32, 24, 1024) 0           conv2d_463[0][0]                 \n",
4177
      "                                                                 conv2d_461[0][0]                 \n",
4178
      "__________________________________________________________________________________________________\n",
4179
      "batch_normalization_324 (BatchN (None, 32, 24, 1024) 4096        add_112[0][0]                    \n",
4180
      "__________________________________________________________________________________________________\n",
4181
      "activation_225 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_324[0][0]    \n",
4182
      "__________________________________________________________________________________________________\n",
4183
      "conv2d_464 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_225[0][0]             \n",
4184
      "__________________________________________________________________________________________________\n",
4185
      "batch_normalization_325 (BatchN (None, 32, 24, 1024) 4096        conv2d_464[0][0]                 \n",
4186
      "__________________________________________________________________________________________________\n",
4187
      "activation_226 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_325[0][0]    \n",
4188
      "__________________________________________________________________________________________________\n",
4189
      "conv2d_465 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_226[0][0]             \n",
4190
      "__________________________________________________________________________________________________\n",
4191
      "add_113 (Add)                   (None, 32, 24, 1024) 0           conv2d_465[0][0]                 \n",
4192
      "                                                                 add_112[0][0]                    \n",
4193
      "__________________________________________________________________________________________________\n",
4194
      "batch_normalization_326 (BatchN (None, 32, 24, 1024) 4096        add_113[0][0]                    \n",
4195
      "__________________________________________________________________________________________________\n",
4196
      "activation_227 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_326[0][0]    \n",
4197
      "__________________________________________________________________________________________________\n",
4198
      "conv2d_466 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_227[0][0]             \n",
4199
      "__________________________________________________________________________________________________\n",
4200
      "batch_normalization_327 (BatchN (None, 32, 24, 1024) 4096        conv2d_466[0][0]                 \n",
4201
      "__________________________________________________________________________________________________\n",
4202
      "activation_228 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_327[0][0]    \n",
4203
      "__________________________________________________________________________________________________\n",
4204
      "conv2d_467 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_228[0][0]             \n",
4205
      "__________________________________________________________________________________________________\n",
4206
      "add_114 (Add)                   (None, 32, 24, 1024) 0           conv2d_467[0][0]                 \n",
4207
      "                                                                 add_113[0][0]                    \n",
4208
      "__________________________________________________________________________________________________\n",
4209
      "batch_normalization_328 (BatchN (None, 32, 24, 1024) 4096        add_114[0][0]                    \n",
4210
      "__________________________________________________________________________________________________\n",
4211
      "activation_229 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_328[0][0]    \n",
4212
      "__________________________________________________________________________________________________\n",
4213
      "conv2d_468 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_229[0][0]             \n",
4214
      "__________________________________________________________________________________________________\n",
4215
      "batch_normalization_329 (BatchN (None, 32, 24, 1024) 4096        conv2d_468[0][0]                 \n",
4216
      "__________________________________________________________________________________________________\n",
4217
      "activation_230 (Activation)     (None, 32, 24, 1024) 0           batch_normalization_329[0][0]    \n",
4218
      "__________________________________________________________________________________________________\n",
4219
      "conv2d_469 (Conv2D)             (None, 32, 24, 1024) 9438208     activation_230[0][0]             \n",
4220
      "__________________________________________________________________________________________________\n",
4221
      "add_115 (Add)                   (None, 32, 24, 1024) 0           conv2d_469[0][0]                 \n",
4222
      "                                                                 add_114[0][0]                    \n",
4223
      "__________________________________________________________________________________________________\n",
4224
      "conv2d_470 (Conv2D)             (None, 32, 24, 512)  524800      add_115[0][0]                    \n",
4225
      "__________________________________________________________________________________________________\n",
4226
      "conv2d_471 (Conv2D)             (None, 32, 24, 570)  2627130     conv2d_470[0][0]                 \n",
4227
      "__________________________________________________________________________________________________\n",
4228
      "conv2d_472 (Conv2D)             (None, 32, 24, 512)  292352      conv2d_471[0][0]                 \n",
4229
      "__________________________________________________________________________________________________\n",
4230
      "conv2d_473 (Conv2D)             (None, 32, 24, 570)  2627130     conv2d_472[0][0]                 \n",
4231
      "__________________________________________________________________________________________________\n",
4232
      "conv2d_474 (Conv2D)             (None, 32, 24, 512)  292352      conv2d_473[0][0]                 \n",
4233
      "__________________________________________________________________________________________________\n",
4234
      "conv2d_475 (Conv2D)             (None, 32, 24, 570)  2627130     conv2d_474[0][0]                 \n",
4235
      "__________________________________________________________________________________________________\n",
4236
      "conv2d_477 (Conv2D)             (None, 32, 24, 256)  146176      conv2d_475[0][0]                 \n",
4237
      "__________________________________________________________________________________________________\n",
4238
      "up_sampling2d_9 (UpSampling2D)  (None, 64, 48, 256)  0           conv2d_477[0][0]                 \n",
4239
      "__________________________________________________________________________________________________\n",
4240
      "concatenate_28 (Concatenate)    (None, 64, 48, 768)  0           add_111[0][0]                    \n",
4241
      "                                                                 up_sampling2d_9[0][0]            \n",
4242
      "__________________________________________________________________________________________________\n",
4243
      "conv2d_478 (Conv2D)             (None, 64, 48, 256)  196864      concatenate_28[0][0]             \n",
4244
      "__________________________________________________________________________________________________\n",
4245
      "conv2d_479 (Conv2D)             (None, 64, 48, 314)  723770      conv2d_478[0][0]                 \n",
4246
      "__________________________________________________________________________________________________\n",
4247
      "conv2d_480 (Conv2D)             (None, 64, 48, 256)  80640       conv2d_479[0][0]                 \n",
4248
      "__________________________________________________________________________________________________\n",
4249
      "conv2d_481 (Conv2D)             (None, 64, 48, 314)  723770      conv2d_480[0][0]                 \n",
4250
      "__________________________________________________________________________________________________\n",
4251
      "conv2d_482 (Conv2D)             (None, 64, 48, 256)  80640       conv2d_481[0][0]                 \n",
4252
      "__________________________________________________________________________________________________\n",
4253
      "conv2d_483 (Conv2D)             (None, 64, 48, 314)  723770      conv2d_482[0][0]                 \n",
4254
      "__________________________________________________________________________________________________\n",
4255
      "conv2d_485 (Conv2D)             (None, 64, 48, 256)  80640       conv2d_483[0][0]                 \n",
4256
      "__________________________________________________________________________________________________\n",
4257
      "up_sampling2d_10 (UpSampling2D) (None, 128, 96, 256) 0           conv2d_485[0][0]                 \n",
4258
      "__________________________________________________________________________________________________\n",
4259
      "concatenate_29 (Concatenate)    (None, 128, 96, 512) 0           add_103[0][0]                    \n",
4260
      "                                                                 up_sampling2d_10[0][0]           \n",
4261
      "__________________________________________________________________________________________________\n",
4262
      "conv2d_486 (Conv2D)             (None, 128, 96, 128) 65664       concatenate_29[0][0]             \n",
4263
      "__________________________________________________________________________________________________\n",
4264
      "conv2d_487 (Conv2D)             (None, 128, 96, 186) 214458      conv2d_486[0][0]                 \n",
4265
      "__________________________________________________________________________________________________\n",
4266
      "conv2d_488 (Conv2D)             (None, 128, 96, 128) 23936       conv2d_487[0][0]                 \n",
4267
      "__________________________________________________________________________________________________\n",
4268
      "conv2d_489 (Conv2D)             (None, 128, 96, 186) 214458      conv2d_488[0][0]                 \n",
4269
      "__________________________________________________________________________________________________\n",
4270
      "conv2d_490 (Conv2D)             (None, 128, 96, 128) 23936       conv2d_489[0][0]                 \n",
4271
      "__________________________________________________________________________________________________\n",
4272
      "conv2d_491 (Conv2D)             (None, 128, 96, 186) 214458      conv2d_490[0][0]                 \n",
4273
      "__________________________________________________________________________________________________\n",
4274
      "conv2d_492 (Conv2D)             (None, 128, 96, 58)  10846       conv2d_491[0][0]                 \n",
4275
      "__________________________________________________________________________________________________\n",
4276
      "conv2d_484 (Conv2D)             (None, 64, 48, 58)   18270       conv2d_483[0][0]                 \n",
4277
      "__________________________________________________________________________________________________\n",
4278
      "conv2d_476 (Conv2D)             (None, 32, 24, 58)   33118       conv2d_475[0][0]                 \n",
4279
      "==================================================================================================\n",
4280
      "Total params: 142,306,468\n",
4281
      "Trainable params: 142,264,228\n",
4282
      "Non-trainable params: 42,240\n",
4283
      "__________________________________________________________________________________________________\n"
4284
     ]
4285
    }
4286
   ],
4287
   "source": [
4288
    "from keras.models import Input, Model\n",
4289
    "from keras.layers import Convolution2D, MaxPooling2D, UpSampling2D, Conv2D, Concatenate, Activation, Dropout\n",
4290
    "from keras.layers import Conv2DTranspose, RepeatVector, ZeroPadding2D, Cropping2D, multiply, average, add, subtract\n",
4291
    "from keras.layers import GlobalAveragePooling2D, GlobalMaxPooling2D, GlobalAveragePooling1D\n",
4292
    "from keras.layers.normalization import BatchNormalization\n",
4293
    "\n",
4294
    "def mySoloV3 (img_shape, num_classes) :\n",
4295
    "    \n",
4296
    "    i = Input(shape=img_shape,   name='input');\n",
4297
    "    \n",
4298
    "    a = i;\n",
4299
    "    \n",
4300
    "    a = convolution_block(a, 32, (3,3),            activation='LeakRelu')\n",
4301
    "    a = convolution_block(a, 64, (3,3), strides=2, activation='LeakRelu')\n",
4302
    "    \n",
4303
    "    a = residual_block  (a, num_filters=64, seOK=False, batch_activate=False)\n",
4304
    "    a = convolution_block(a, 128, (3,3), strides=2, activation='LeakRelu')\n",
4305
    "    \n",
4306
    "    for ii in range(2) :\n",
4307
    "        a = residual_block  (a, num_filters=128, seOK=False, batch_activate=False)\n",
4308
    "    \n",
4309
    "    a = convolution_block(a, 256, (3,3), strides=2, activation='LeakRelu')\n",
4310
    "\n",
4311
    "    for ii in range(8) :\n",
4312
    "        a = residual_block  (a, num_filters=256, seOK=False, batch_activate=False)\n",
4313
    "        \n",
4314
    "        \n",
4315
    "    y1 = a\n",
4316
    "    \n",
4317
    "    a = convolution_block(a, 512, (3,3), strides=2, activation='LeakRelu')\n",
4318
    "\n",
4319
    "    for ii in range(8) :\n",
4320
    "        a = residual_block  (a, num_filters=512, seOK=False, batch_activate=False)\n",
4321
    "    \n",
4322
    "    y2 = a\n",
4323
    "    \n",
4324
    "    a = convolution_block(a, 1024, (3,3), strides=2, activation='LeakRelu')\n",
4325
    "    \n",
4326
    "\n",
4327
    "    for ii in range(4) :\n",
4328
    "        a = residual_block  (a, num_filters=1024, seOK=False, batch_activate=False)\n",
4329
    "    \n",
4330
    "    y3 = a\n",
4331
    "    \n",
4332
    "    for _ in range(3) : \n",
4333
    "        y3 = convolution_block(y3, 512,  (1,1), activation=False)\n",
4334
    "        y3 = convolution_block(y3, 512+(5+num_classes), (3,3), activation=False)  \n",
4335
    "        \n",
4336
    "    y23= y3\n",
4337
    "    y3 = convolution_block(y3, (5+num_classes), (1,1), activation=False)    \n",
4338
    "\n",
4339
    "\n",
4340
    "    y23= convolution_block(y23, 256, (1,1), activation=False)\n",
4341
    "    y23= UpSampling2D((2,2))(y23)\n",
4342
    "    y2 = Concatenate()([y2,y23])\n",
4343
    "    \n",
4344
    "    for _ in range(3) : \n",
4345
    "        y2 = convolution_block(y2, 256,  (1,1), activation=False)\n",
4346
    "        y2 = convolution_block(y2, 256+(5+num_classes), (3,3), activation=False)  \n",
4347
    "        \n",
4348
    "    y12= y2\n",
4349
    "    y2 = convolution_block(y2, (5+num_classes), (1,1), activation=False)    \n",
4350
    "\n",
4351
    "    y12= convolution_block(y12, 256, (1,1), activation=False)\n",
4352
    "    y12= UpSampling2D((2,2))(y12)\n",
4353
    "    y1 = Concatenate()([y1,y12])\n",
4354
    "    \n",
4355
    "    for _ in range(3) : \n",
4356
    "        y1 = convolution_block(y1, 128,  (1,1), activation=False)\n",
4357
    "        y1 = convolution_block(y1, 128+(5+num_classes), (3,3), activation=False)  \n",
4358
    "        \n",
4359
    "    y1 = convolution_block(y1, (5+num_classes), (1,1), activation=False)    \n",
4360
    "\n",
4361
    "    \n",
4362
    "    #o = Concatenate()([y1,y2,y3])\n",
4363
    "        \n",
4364
    "\n",
4365
    "    return Model(inputs=i, outputs=[y1,y2,y3], name='mySoloV3')\n",
4366
    "\n",
4367
    "if 1 :\n",
4368
    "\n",
4369
    "    model10 = mySoloV3((1024,512+256,1), num_classes=53)\n",
4370
    "    model10.summary()    "
4371
   ]
4372
  },
4373
  {
4374
   "cell_type": "code",
4375
   "execution_count": null,
4376
   "metadata": {},
4377
   "outputs": [],
4378
   "source": []
4379
  }
4380
 ],
4381
 "metadata": {
4382
  "kernelspec": {
4383
   "display_name": "Python 3",
4384
   "language": "python",
4385
   "name": "python3"
4386
  },
4387
  "language_info": {
4388
   "codemirror_mode": {
4389
    "name": "ipython",
4390
    "version": 3
4391
   },
4392
   "file_extension": ".py",
4393
   "mimetype": "text/x-python",
4394
   "name": "python",
4395
   "nbconvert_exporter": "python",
4396
   "pygments_lexer": "ipython3",
4397
   "version": "3.6.6"
4398
  }
4399
 },
4400
 "nbformat": 4,
4401
 "nbformat_minor": 2
4402
}
4403