-
Notifications
You must be signed in to change notification settings - Fork 125
/
cosin_add_m_layer.cpp
124 lines (107 loc) · 4.02 KB
/
cosin_add_m_layer.cpp
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
#include <algorithm>
#include <vector>
#include <math.h>
#include "caffe/layers/cosin_add_m_layer.hpp"
namespace caffe {
template <typename Dtype>
void CosinAddmLayer<Dtype>::LayerSetUp(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
const CosinAddmParameter& param = this->layer_param_.cosin_add_m_param();
m_ = param.m();
sin_m = sin(m_);
cos_m = cos(m_);
threshold = cos(M_PI - m_);
count_num = 0;
count_num_back = 0;
transform_test_ = param.transform_test() & (this->phase_ == TRAIN);
}
template <typename Dtype>
void CosinAddmLayer<Dtype>::Reshape(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
top[0]->ReshapeLike(*bottom[0]);
top_flag.ReshapeLike(*bottom[0]);
cos_theta.ReshapeLike(*bottom[0]);
}
template <typename Dtype>
void CosinAddmLayer<Dtype>::Forward_cpu(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
const Dtype* bottom_data = bottom[0]->cpu_data();
const Dtype* label_data = bottom[1]->cpu_data();
Dtype* top_data = top[0]->mutable_cpu_data();
Dtype* tpflag = top_flag.mutable_cpu_data();
Dtype* cos_t = cos_theta.mutable_cpu_data();
int num = bottom[0]->num();
int count = bottom[0]->count();
int dim = count / num;
caffe_copy(count, bottom_data, top_data);
caffe_copy(count, bottom_data, cos_t);
caffe_set(count, Dtype(0), tpflag);
for (int i = 0; i < num; ++i) {
int gt = static_cast<int>(label_data[i]);
Dtype sin_theta_2 = cos_t[i * dim + gt] * cos_t[i * dim + gt];
Dtype sin_theta = sqrt(1.0f - sin_theta_2);
if(cos_t[i * dim + gt] > 1.0f)
{
LOG(INFO) << "cos_theta > 1 ****** " << cos_t[i * dim + gt];
cos_t[i * dim + gt] = 1.0f;
sin_theta_2 = 1.0f;
sin_theta = 0.0f;
}
if(cos_t[i * dim + gt] <= threshold)
{
top_data[i * dim + gt] = cos_t[i * dim + gt] - sin(M_PI - m_) * m_;
tpflag[i * dim + gt] = 1.0f;
}
else
top_data[i * dim + gt] = cos_t[i * dim + gt] * cos_m - sin_theta * sin_m;
// if(count_num % 10 == 0)
// {
// LOG(INFO) << "top_data[" << i * dim + gt << "]: " << top_data[i * dim + gt] << " cos_t: " << cos_t[i * dim + gt];
// }
}
count_num ++;
}
template <typename Dtype>
void CosinAddmLayer<Dtype>::Backward_cpu(const vector<Blob<Dtype>*>& top,
const vector<bool>& propagate_down,
const vector<Blob<Dtype>*>& bottom) {
if (propagate_down[0]) {
const Dtype* top_diff = top[0]->cpu_diff();
const Dtype* label_data = bottom[1]->cpu_data();
const Dtype* cos_t = cos_theta.cpu_data();
Dtype* bottom_diff = bottom[0]->mutable_cpu_diff();
const Dtype* tpflag = top_flag.cpu_data();
int count = bottom[0]->count();
caffe_copy(count, top_diff, bottom_diff);
int num = bottom[0]->num();
int dim = count / num;
for (int i = 0; i < num; ++i)
{
int gt = static_cast<int>(label_data[i]);
Dtype sin_theta_2 = cos_t[i * dim + gt] * cos_t[i * dim + gt];
if(cos_t[i * dim + gt] == 1.0f)
{
sin_theta_2 = 1.0f;
}
Dtype sin_theta = sqrt(1.0f - sin_theta_2);
Dtype coffe = 0.0f;
if(sin_theta == 0.0f)
coffe = 1.0f;
else
coffe = cos_m + sin_m * cos_t[i * dim + gt] / sin_theta;
if(tpflag[i * dim + gt] > 0.0f)
coffe = 1.0f;
bottom_diff[i * dim + gt] = coffe * top_diff[i * dim + gt];
// if(count_num_back % 10 == 0)
// {
// LOG(INFO) << "top_diff: " << top_diff[i * dim + gt];
// LOG(INFO) << "bottom_diff: " << bottom_diff[i * dim + gt];
// LOG(INFO) << "cos_theta[ "<<i * dim + gt <<"]: "<< cos_t[i * dim + gt]<< " coffe: " << coffe;
// }
}
}
count_num_back++;
}
INSTANTIATE_CLASS(CosinAddmLayer);
REGISTER_LAYER_CLASS(CosinAddm);
} // namespace caffe