ncnn

1
// Tencent is pleased to support the open source community by making ncnn available.
2
//
3
// Copyright (C) 2020 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_priorbox_caffe()
18
{
19
    ncnn::Mat min_sizes(1);
20
    min_sizes[0] = 105.f;
21

22
    ncnn::Mat max_sizes(1);
23
    max_sizes[0] = 150.f;
24

25
    ncnn::Mat aspect_ratios(2);
26
    aspect_ratios[0] = 2.f;
27
    aspect_ratios[1] = 3.f;
28

29
    ncnn::ParamDict pd;
30
    pd.set(0, min_sizes);
31
    pd.set(1, max_sizes);
32
    pd.set(2, aspect_ratios);
33
    pd.set(3, 0.1f);    // variances[0]
34
    pd.set(4, 0.1f);    // variances[1]
35
    pd.set(5, 0.2f);    // variances[2]
36
    pd.set(6, 0.2f);    // variances[3]
37
    pd.set(7, 1);       // flip
38
    pd.set(8, 0);       // clip
39
    pd.set(9, -233);    // image_width
40
    pd.set(10, -233);   // image_height
41
    pd.set(11, -233.f); // step_width
42
    pd.set(12, -233.f); // step_height
43
    pd.set(13, 0.f);    // offset
44
    pd.set(14, 0.f);    // step_mmdetection
45
    pd.set(15, 0.f);    // center_mmdetection
46

47
    std::vector<ncnn::Mat> weights(0);
48

49
    std::vector<ncnn::Mat> as(2);
50
    as[0] = RandomMat(72, 72, 1);
51
    as[1] = RandomMat(512, 512, 1);
52

53
    int ret = test_layer("PriorBox", pd, weights, as, 1);
54
    if (ret != 0)
55
    {
56
        fprintf(stderr, "test_priorbox_caffe failed\n");
57
    }
58

59
    return ret;
60
}
61

62
static int test_priorbox_mxnet()
63
{
64
    ncnn::Mat min_sizes(2);
65
    min_sizes[0] = 0.15f;
66
    min_sizes[1] = 0.2121f;
67

68
    ncnn::Mat max_sizes(0);
69

70
    ncnn::Mat aspect_ratios(5);
71
    aspect_ratios[0] = 1.f;
72
    aspect_ratios[1] = 2.f;
73
    aspect_ratios[2] = 0.5f;
74
    aspect_ratios[3] = 3.f;
75
    aspect_ratios[4] = 0.333333;
76

77
    ncnn::ParamDict pd;
78
    pd.set(0, min_sizes);
79
    pd.set(1, max_sizes);
80
    pd.set(2, aspect_ratios);
81
    pd.set(3, 0.1f);    // variances[0]
82
    pd.set(4, 0.1f);    // variances[1]
83
    pd.set(5, 0.2f);    // variances[2]
84
    pd.set(6, 0.2f);    // variances[3]
85
    pd.set(7, 0);       // flip
86
    pd.set(8, 0);       // clip
87
    pd.set(9, -233);    // image_width
88
    pd.set(10, -233);   // image_height
89
    pd.set(11, -233.f); // step_width
90
    pd.set(12, -233.f); // step_height
91
    pd.set(13, 0.5f);   // offset
92
    pd.set(14, 0.f);    // step_mmdetection
93
    pd.set(15, 0.f);    // center_mmdetection
94

95
    std::vector<ncnn::Mat> weights(0);
96

97
    std::vector<ncnn::Mat> as(1);
98
    as[0] = RandomMat(72, 72, 1);
99

100
    int ret = test_layer("PriorBox", pd, weights, as, 1);
101
    if (ret != 0)
102
    {
103
        fprintf(stderr, "test_priorbox_mxnet failed\n");
104
    }
105

106
    return ret;
107
}
108

109
int main()
110
{
111
    SRAND(7767517);
112

113
    return 0
114
           || test_priorbox_caffe()
115
           || test_priorbox_mxnet();
116
}
117