【Caffe代码解析】Layer网络层

#ifndef CAFFE_LAYER_H_
#define CAFFE_LAYER_H_

#include <algorithm>
#include <string>
#include <vector>

#include "caffe/blob.hpp"
#include "caffe/common.hpp"
#include "caffe/layer_factory.hpp"
#include "caffe/proto/caffe.pb.h"
#include "caffe/util/device_alternate.hpp"

namespace caffe {
// 功能：全部的网络层的基类，定义的全部的网络层的通用接口。

// 前馈接口，必须实现
// 反馈接口，须要的时候实现，计算梯度。

template <typename Dtype>
class Layer {
public:
/**
* 每个网络层须要自定义它的setup而不须要构造函数
*/
explicit Layer(const LayerParameter& param)
: layer_param_(param) {
//通过网络层參数来构造网络层
phase_ = param.phase();
if (layer_param_.blobs_size() > 0) {
blobs_.resize(layer_param_.blobs_size());
for (int i = 0; i < layer_param_.blobs_size(); ++i) {
blobs_[i].reset(new Blob<Dtype>());
blobs_[i]->FromProto(layer_param_.blobs(i));
}
}
}
// 析构函数
virtual ~Layer() {}

/**
* 实现一些通用的设置功能
*
* @param bottom 网络层的输入的shape
* @param top 网络层的输出，须要被reshape
* 调用 LayerSetUp 来对每个网络层进行特殊化的处理,
* 调用reshape top
* 设置 数值权重
* 这种方法能够不被重载。

*/
void SetUp(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
CheckBlobCounts(bottom, top);
LayerSetUp(bottom, top);
Reshape(bottom, top);
SetLossWeights(top);
}

/**
* @brief 设置一些层相关的设置，定义的层须要实现这种方法以及Reshape方法
*/
//设置网络层
virtual void LayerSetUp(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {}
/**
* @brief 调整top blob以适应bottom blob。
*/
virtual void Reshape(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) = 0;

/**
* @brief 给定 bottom blobs, 计算 top blobs 以及 loss.
* 每个网络层都须要定义cpu版本号的前馈，可选gpu版本号的前馈
*/
//前馈
inline Dtype Forward(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top);

/**
* @brief 给定 top blob 的梯度, 计算 bottom blob 梯度.
* @param propagate_down 向量，长度为ibottom   的个数。每个索引值表示是是否将损失梯度值反馈到该bottom中
*/
//反馈
inline void Backward(const vector<Blob<Dtype>*>& top,
const vector<bool>& propagate_down,
const vector<Blob<Dtype>*>& bottom);

/**
* @brief 返回可学习的參数 blobs.
*/
vector<shared_ptr<Blob<Dtype> > >& blobs() {
return blobs_;
}

/**
* @brief 返回网络层參数
*/
const LayerParameter& layer_param() const { return layer_param_; }
//序列化
virtual void ToProto(LayerParameter* param, bool write_diff = false);

/**
* @brief 返回指定索引的标量损失值。
*/
inline Dtype loss(const int top_index) const {
return (loss_.size() > top_index) ?

loss_[top_index] : Dtype(0);
}

/**
* @brief 设置网络层制定索引位置的loss
*/
inline void set_loss(const int top_index, const Dtype value) {
if (loss_.size() <= top_index) {
loss_.resize(top_index + 1, Dtype(0));
}
loss_[top_index] = value;
}

/**
* @brief 返回网络层名字，字符串描写叙述u
*/
virtual inline const char* type() const { return ""; }
//Bottom的blob的确切数目
virtual inline int ExactNumBottomBlobs() const { return -1; }
//Bottom blob的最小数目
virtual inline int MinBottomBlobs() const { return -1; }
//Botttom的确切数目
virtual inline int MaxBottomBlobs() const { return -1; }
//Top Blob的确切数目
virtual inline int ExactNumTopBlobs() const { return -1; }
//最小的blob的数目
virtual inline int MinTopBlobs() const { return -1; }
// 最大的blob的数目
virtual inline int MaxTopBlobs() const { return -1; }
// 是否bottom 和top的数目同样
virtual inline bool EqualNumBottomTopBlobs() const { return false; }
// 是否自己主动Top blob
virtual inline bool AutoTopBlobs() const { return false; }
//查询某一个bottom是否强制bp
virtual inline bool AllowForceBackward(const int bottom_index) const {
return true;
}
//查询某一个blob是否bp
inline bool param_propagate_down(const int param_id) {
return (param_propagate_down_.size() > param_id) ?
param_propagate_down_[param_id] : false;
}
//设置某一个blob是否bp。
inline void set_param_propagate_down(const int param_id, const bool value) {
if (param_propagate_down_.size() <= param_id) {
param_propagate_down_.resize(param_id + 1, true);
}
param_propagate_down_[param_id] = value;
}

protected:
// 网络层參数
LayerParameter layer_param_;
// 模式
Phase phase_;
//用blob来存储一系列向量
vector<shared_ptr<Blob<Dtype> > > blobs_;
//是否bp的向量
vector<bool> param_propagate_down_;
//存储top的loss
vector<Dtype> loss_;
//cpu版本号的前馈实现
virtual void Forward_cpu(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) = 0;
//gpu版本号的前馈实现
virtual void Forward_gpu(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
// LOG(WARNING) << "Using CPU code as backup.";
return Forward_cpu(bottom, top);
}
//cpu版本号的前馈实现
virtual void Backward_cpu(const vector<Blob<Dtype>*>& top,
const vector<bool>& propagate_down,
const vector<Blob<Dtype>*>& bottom) = 0;
//gpu版本号的反馈实现
virtual void Backward_gpu(const vector<Blob<Dtype>*>& top,
const vector<bool>& propagate_down,
const vector<Blob<Dtype>*>& bottom) {
// LOG(WARNING) << "Using CPU code as backup.";
Backward_cpu(top, propagate_down, bottom);
}
// 核查bootom和top的大小是否与该layer层指定的一致。

virtual void CheckBlobCounts(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
if (ExactNumBottomBlobs() >= 0) {
CHECK_EQ(ExactNumBottomBlobs(), bottom.size())
<< type() << " Layer takes " << ExactNumBottomBlobs()
<< " bottom blob(s) as input.";
}
if (MinBottomBlobs() >= 0) {
CHECK_LE(MinBottomBlobs(), bottom.size())
<< type() << " Layer takes at least " << MinBottomBlobs()
<< " bottom blob(s) as input.";
}
if (MaxBottomBlobs() >= 0) {
CHECK_GE(MaxBottomBlobs(), bottom.size())
<< type() << " Layer takes at most " << MaxBottomBlobs()
<< " bottom blob(s) as input.";
}
if (ExactNumTopBlobs() >= 0) {
CHECK_EQ(ExactNumTopBlobs(), top.size())
<< type() << " Layer produces " << ExactNumTopBlobs()
<< " top blob(s) as output.";
}
if (MinTopBlobs() >= 0) {
CHECK_LE(MinTopBlobs(), top.size())
<< type() << " Layer produces at least " << MinTopBlobs()
<< " top blob(s) as output.";
}
if (MaxTopBlobs() >= 0) {
CHECK_GE(MaxTopBlobs(), top.size())
<< type() << " Layer produces at most " << MaxTopBlobs()
<< " top blob(s) as output.";
}
if (EqualNumBottomTopBlobs()) {
CHECK_EQ(bottom.size(), top.size())
<< type() << " Layer produces one top blob as output for each "
<< "bottom blob input.";
}
}
// 用blob初始化损失权重。
inline void SetLossWeights(const vector<Blob<Dtype>*>& top) {
const int num_loss_weights = layer_param_.loss_weight_size();
if (num_loss_weights) {
CHECK_EQ(top.size(), num_loss_weights) << "loss_weight must be "
"unspecified or specified once per top blob.";
for (int top_id = 0; top_id < top.size(); ++top_id) {
const Dtype loss_weight = layer_param_.loss_weight(top_id);
if (loss_weight == Dtype(0)) { continue; }
this->set_loss(top_id, loss_weight);
const int count = top[top_id]->count();
Dtype* loss_multiplier = top[top_id]->mutable_cpu_diff();
caffe_set(count, loss_weight, loss_multiplier);
}
}
}

DISABLE_COPY_AND_ASSIGN(Layer);
};  // class Layer

// 前馈。依据caffe的mode 调用相相应的cpu实现或者是gpu实现。而且计算损失函数值。

template <typename Dtype>
inline Dtype Layer<Dtype>::Forward(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
Dtype loss = 0;
Reshape(bottom, top);
switch (Caffe::mode()) {
case Caffe::CPU:
Forward_cpu(bottom, top);
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->cpu_data();
const Dtype* loss_weights = top[top_id]->cpu_diff();
loss += caffe_cpu_dot(count, data, loss_weights);
}
break;
case Caffe::GPU:
Forward_gpu(bottom, top);
#ifndef CPU_ONLY
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->gpu_data();
const Dtype* loss_weights = top[top_id]->gpu_diff();
Dtype blob_loss = 0;
caffe_gpu_dot(count, data, loss_weights, &blob_loss);
loss += blob_loss;
}
#endif
break;
default:
LOG(FATAL) << "Unknown caffe mode.";
}
return loss;
}

//反向传播梯度。依据Caffe的mode是在GPU还是CPU，调用相相应版本号的函数
//propagate_down 用于控制相应的层是否bp
template <typename Dtype>
inline void Layer<Dtype>::Backward(const vector<Blob<Dtype>*>& top,
const vector<bool>& propagate_down,
const vector<Blob<Dtype>*>& bottom) {
switch (Caffe::mode()) {
case Caffe::CPU:
Backward_cpu(top, propagate_down, bottom);
break;
case Caffe::GPU:
Backward_gpu(top, propagate_down, bottom);
break;
default:
LOG(FATAL) << "Unknown caffe mode.";
}
}

// 序列化网络层參数到协议缓存。终于是调用blob写入协议缓存。
template <typename Dtype>
void Layer<Dtype>::ToProto(LayerParameter* param, bool write_diff) {
param->Clear();
param->CopyFrom(layer_param_);
param->clear_blobs();
for (int i = 0; i < blobs_.size(); ++i) {
blobs_[i]->ToProto(param->add_blobs(), write_diff);
}
}

}  // namespace caffe

#endif  // CAFFE_LAYER_H_