1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
|
#include <algorithm>
#include <cmath>
#include <vector>
#include "gtest/gtest.h"
#include "caffe/blob.hpp"
#include "caffe/common.hpp"
#include "caffe/filler.hpp"
#include "caffe/layers/contrastive_loss_layer.hpp"
#include "caffe/test/test_caffe_main.hpp"
#include "caffe/test/test_gradient_check_util.hpp"
namespace caffe {
template <typename TypeParam>
class ContrastiveLossLayerTest : public MultiDeviceTest<TypeParam> {
typedef typename TypeParam::Dtype Dtype;
protected:
ContrastiveLossLayerTest()
: blob_bottom_data_i_(new Blob<Dtype>(512, 2, 1, 1)),
blob_bottom_data_j_(new Blob<Dtype>(512, 2, 1, 1)),
blob_bottom_y_(new Blob<Dtype>(512, 1, 1, 1)),
blob_top_loss_(new Blob<Dtype>()) {
// fill the values
FillerParameter filler_param;
filler_param.set_min(-1.0);
filler_param.set_max(1.0); // distances~=1.0 to test both sides of margin
UniformFiller<Dtype> filler(filler_param);
filler.Fill(this->blob_bottom_data_i_);
blob_bottom_vec_.push_back(blob_bottom_data_i_);
filler.Fill(this->blob_bottom_data_j_);
blob_bottom_vec_.push_back(blob_bottom_data_j_);
for (int i = 0; i < blob_bottom_y_->count(); ++i) {
blob_bottom_y_->mutable_cpu_data()[i] = caffe_rng_rand() % 2; // 0 or 1
}
blob_bottom_vec_.push_back(blob_bottom_y_);
blob_top_vec_.push_back(blob_top_loss_);
}
virtual ~ContrastiveLossLayerTest() {
delete blob_bottom_data_i_;
delete blob_bottom_data_j_;
delete blob_bottom_y_;
delete blob_top_loss_;
}
Blob<Dtype>* const blob_bottom_data_i_;
Blob<Dtype>* const blob_bottom_data_j_;
Blob<Dtype>* const blob_bottom_y_;
Blob<Dtype>* const blob_top_loss_;
vector<Blob<Dtype>*> blob_bottom_vec_;
vector<Blob<Dtype>*> blob_top_vec_;
};
TYPED_TEST_CASE(ContrastiveLossLayerTest, TestDtypesAndDevices);
TYPED_TEST(ContrastiveLossLayerTest, TestForward) {
typedef typename TypeParam::Dtype Dtype;
LayerParameter layer_param;
ContrastiveLossLayer<Dtype> layer(layer_param);
layer.SetUp(this->blob_bottom_vec_, this->blob_top_vec_);
layer.Forward(this->blob_bottom_vec_, this->blob_top_vec_);
// manually compute to compare
const Dtype margin = layer_param.contrastive_loss_param().margin();
const int num = this->blob_bottom_data_i_->num();
const int channels = this->blob_bottom_data_i_->channels();
Dtype loss(0);
for (int i = 0; i < num; ++i) {
Dtype dist_sq(0);
for (int j = 0; j < channels; ++j) {
Dtype diff = this->blob_bottom_data_i_->cpu_data()[i*channels+j] -
this->blob_bottom_data_j_->cpu_data()[i*channels+j];
dist_sq += diff*diff;
}
if (this->blob_bottom_y_->cpu_data()[i]) { // similar pairs
loss += dist_sq;
} else {
Dtype dist = std::max<Dtype>(margin - sqrt(dist_sq), 0.0);
loss += dist*dist;
}
}
loss /= static_cast<Dtype>(num) * Dtype(2);
EXPECT_NEAR(this->blob_top_loss_->cpu_data()[0], loss, 1e-6);
}
TYPED_TEST(ContrastiveLossLayerTest, TestGradient) {
typedef typename TypeParam::Dtype Dtype;
LayerParameter layer_param;
ContrastiveLossLayer<Dtype> layer(layer_param);
layer.SetUp(this->blob_bottom_vec_, this->blob_top_vec_);
GradientChecker<Dtype> checker(1e-2, 1e-2, 1701);
// check the gradient for the first two bottom layers
checker.CheckGradientExhaustive(&layer, this->blob_bottom_vec_,
this->blob_top_vec_, 0);
checker.CheckGradientExhaustive(&layer, this->blob_bottom_vec_,
this->blob_top_vec_, 1);
}
TYPED_TEST(ContrastiveLossLayerTest, TestForwardLegacy) {
typedef typename TypeParam::Dtype Dtype;
LayerParameter layer_param;
layer_param.mutable_contrastive_loss_param()->set_legacy_version(true);
ContrastiveLossLayer<Dtype> layer(layer_param);
layer.SetUp(this->blob_bottom_vec_, this->blob_top_vec_);
layer.Forward(this->blob_bottom_vec_, this->blob_top_vec_);
// manually compute to compare
const Dtype margin = layer_param.contrastive_loss_param().margin();
const int num = this->blob_bottom_data_i_->num();
const int channels = this->blob_bottom_data_i_->channels();
Dtype loss(0);
for (int i = 0; i < num; ++i) {
Dtype dist_sq(0);
for (int j = 0; j < channels; ++j) {
Dtype diff = this->blob_bottom_data_i_->cpu_data()[i*channels+j] -
this->blob_bottom_data_j_->cpu_data()[i*channels+j];
dist_sq += diff*diff;
}
if (this->blob_bottom_y_->cpu_data()[i]) { // similar pairs
loss += dist_sq;
} else {
loss += std::max(margin - dist_sq, Dtype(0.0));
}
}
loss /= static_cast<Dtype>(num) * Dtype(2);
EXPECT_NEAR(this->blob_top_loss_->cpu_data()[0], loss, 1e-6);
}
TYPED_TEST(ContrastiveLossLayerTest, TestGradientLegacy) {
typedef typename TypeParam::Dtype Dtype;
LayerParameter layer_param;
layer_param.mutable_contrastive_loss_param()->set_legacy_version(true);
ContrastiveLossLayer<Dtype> layer(layer_param);
layer.SetUp(this->blob_bottom_vec_, this->blob_top_vec_);
GradientChecker<Dtype> checker(1e-2, 1e-2, 1701);
// check the gradient for the first two bottom layers
checker.CheckGradientExhaustive(&layer, this->blob_bottom_vec_,
this->blob_top_vec_, 0);
checker.CheckGradientExhaustive(&layer, this->blob_bottom_vec_,
this->blob_top_vec_, 1);
}
} // namespace caffe
|
