ncnn

test_convolutiondepthwise.cpp
132 строки · 5.8 Кб
Перенос по словам
1
// Tencent is pleased to support the open source community by making ncnn available.
2
//
3
// Copyright (C) 2019 THL A29 Limited, a Tencent company. All rights reserved.
4
//
5
// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
6
// in compliance with the License. You may obtain a copy of the License at
7
//
8
// https://opensource.org/licenses/BSD-3-Clause
9
//
10
// Unless required by applicable law or agreed to in writing, software distributed
11
// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
12
// CONDITIONS OF ANY KIND, either express or implied. See the License for the
13
// specific language governing permissions and limitations under the License.
14

15
#include "testutil.h"
16

17
static int test_convolutiondepthwise(int w, int h, int c, int outch, int kernel, int dilation, int stride, int pad, int bias, int group)
18
{
19
    ncnn::Mat a = RandomMat(w, h, c);
20

21
    ncnn::ParamDict pd;
22
    pd.set(0, outch);
23
    pd.set(1, kernel);
24
    pd.set(2, dilation);
25
    pd.set(3, stride);
26
    pd.set(4, pad);
27
    pd.set(5, bias);
28
    pd.set(6, outch / group * c / group * kernel * kernel * group);
29
    pd.set(7, group);
30

31
    int activation_type = RAND() % 7; // 0 1 2 3 4 5 6
32
    ncnn::Mat activation_params(2);
33
    activation_params[0] = (activation_type == 6) ? RandomFloat(0, 1) : RandomFloat(-1, 0); // alpha
34
    activation_params[1] = RandomFloat(0, 1);                                               // beta
35
    pd.set(9, activation_type);
36
    pd.set(10, activation_params);
37

38
    std::vector<ncnn::Mat> weights(2);
39
    weights[0] = RandomMat(outch / group * c / group * kernel * kernel * group);
40
    weights[1] = RandomMat(outch);
41

42
    int ret = test_layer("ConvolutionDepthWise", pd, weights, a);
43
    if (ret != 0)
44
    {
45
        fprintf(stderr, "test_convolutiondepthwise failed w=%d h=%d c=%d outch=%d kernel=%d dilation=%d stride=%d pad=%d bias=%d group=%d act=%d actparams=[%f,%f]\n", w, h, c, outch, kernel, dilation, stride, pad, bias, group, activation_type, activation_params[0], activation_params[1]);
46
    }
47

48
    return ret;
49
}
50

51
static int test_convolutiondepthwise_0()
52
{
53
    static const int kdsp[16][4] = {
54
        {1, 1, 1, 0},
55
        {1, 1, 2, 0},
56
        {2, 1, 1, 1},
57
        {2, 1, 2, -233},
58
        {3, 1, 1, 1},
59
        {3, 1, 2, 1},
60
        {3, 2, 1, 1},
61
        {4, 1, 1, 2},
62
        {4, 1, 2, -233},
63
        {4, 2, 1, -234},
64
        {5, 1, 1, -234},
65
        {5, 1, 2, 2},
66
        {5, 2, 2, 2},
67
        {7, 1, 1, 3},
68
        {7, 1, 2, 3},
69
        {7, 2, 1, -233},
70
    };
71

72
    for (int i = 0; i < 16; i++)
73
    {
74
        const int k = kdsp[i][0];
75
        const int d = kdsp[i][1];
76
        const int s = kdsp[i][2];
77
        const int p = kdsp[i][3];
78

79
        int ret = 0
80
                  || test_convolutiondepthwise(15, 7, 1, 1, k, d, s, p, 1, 1)
81
                  || test_convolutiondepthwise(15, 7, 2, 2, k, d, s, p, 0, 1)
82
                  || test_convolutiondepthwise(15, 7, 2, 2, k, d, s, p, 1, 2)
83
                  || test_convolutiondepthwise(15, 7, 3, 3, k, d, s, p, 0, 3)
84
                  || test_convolutiondepthwise(15, 7, 4, 2, k, d, s, p, 1, 2)
85
                  || test_convolutiondepthwise(15, 7, 4, 4, k, d, s, p, 0, 4)
86
                  || test_convolutiondepthwise(15, 7, 7, 7, k, d, s, p, 1, 7)
87
                  || test_convolutiondepthwise(15, 7, 8, 8, k, d, s, p, 0, 2)
88
                  || test_convolutiondepthwise(15, 7, 8, 8, k, d, s, p, 1, 8)
89
                  || test_convolutiondepthwise(15, 7, 12, 12, k, d, s, p, 0, 4)
90
                  || test_convolutiondepthwise(15, 7, 15, 15, k, d, s, p, 1, 15)
91
                  || test_convolutiondepthwise(15, 7, 16, 8, k, d, s, p, 0, 2)
92
                  || test_convolutiondepthwise(15, 7, 16, 16, k, d, s, p, 1, 16)
93
                  || test_convolutiondepthwise(18, 17, 1, 1, k, d, s, p, 1, 1)
94
                  || test_convolutiondepthwise(18, 17, 2, 2, k, d, s, p, 0, 1)
95
                  || test_convolutiondepthwise(18, 17, 2, 2, k, d, s, p, 1, 2)
96
                  || test_convolutiondepthwise(18, 17, 3, 3, k, d, s, p, 0, 3)
97
                  || test_convolutiondepthwise(18, 17, 4, 2, k, d, s, p, 1, 2)
98
                  || test_convolutiondepthwise(18, 17, 4, 4, k, d, s, p, 0, 4)
99
                  || test_convolutiondepthwise(18, 17, 7, 7, k, d, s, p, 1, 7)
100
                  || test_convolutiondepthwise(18, 17, 8, 8, k, d, s, p, 0, 2)
101
                  || test_convolutiondepthwise(18, 17, 8, 8, k, d, s, p, 1, 8)
102
                  || test_convolutiondepthwise(18, 17, 12, 12, k, d, s, p, 0, 4)
103
                  || test_convolutiondepthwise(18, 17, 15, 15, k, d, s, p, 1, 15)
104
                  || test_convolutiondepthwise(18, 17, 16, 8, k, d, s, p, 0, 2)
105
                  || test_convolutiondepthwise(18, 17, 16, 16, k, d, s, p, 1, 16)
106
                  || test_convolutiondepthwise(25, 33, 1, 1, k, d, s, p, 1, 1)
107
                  || test_convolutiondepthwise(25, 33, 2, 2, k, d, s, p, 0, 1)
108
                  || test_convolutiondepthwise(25, 33, 2, 2, k, d, s, p, 1, 2)
109
                  || test_convolutiondepthwise(25, 33, 3, 3, k, d, s, p, 0, 3)
110
                  || test_convolutiondepthwise(25, 33, 4, 2, k, d, s, p, 1, 2)
111
                  || test_convolutiondepthwise(25, 33, 4, 4, k, d, s, p, 0, 4)
112
                  || test_convolutiondepthwise(25, 33, 7, 7, k, d, s, p, 1, 7)
113
                  || test_convolutiondepthwise(25, 33, 8, 8, k, d, s, p, 0, 2)
114
                  || test_convolutiondepthwise(25, 33, 8, 8, k, d, s, p, 1, 8)
115
                  || test_convolutiondepthwise(25, 33, 12, 12, k, d, s, p, 0, 4)
116
                  || test_convolutiondepthwise(25, 33, 15, 15, k, d, s, p, 1, 15)
117
                  || test_convolutiondepthwise(25, 33, 16, 8, k, d, s, p, 0, 2)
118
                  || test_convolutiondepthwise(25, 33, 16, 16, k, d, s, p, 1, 16);
119

120
        if (ret != 0)
121
            return -1;
122
    }
123

124
    return 0;
125
}
126

127
int main()
128
{
129
    SRAND(7767517);
130

131
    return test_convolutiondepthwise_0();
132
}
133
ncnn

Использование cookies
