caffe源码中已经帮我封装好了各种各样的layer,但是有时候现有的layer不能满足设计的网络要求,这个时候需要自己定义一个新的layer,本文参考here,进行简单讲解,具体方式如下:
1.添加你的layer头文件置于 include/caffe/layers/ 下,比如include/caffe/layers/your_layer.hpp
2.your_layer继承选择继承layer.hpp, common_layers.hpp, data_layers.hpp, loss_layers.hpp, neuron_layers.hpp, 或者 vision_layers.hpp其中一种
3.重写
virtual inline const char* type() const { return "YourLayerName"; }
函数,这个的目的是为了在写net.prototxt时,layer{type:"YourLayerName"}
有所对应4.根据自己layer的需要,对{*}blob部分方法进行重写,以此来限制bottom和top的blob个数。比如 要是重写了
virtual inline int ExactNumBottomBlobs() const { return 1; }
就表示限制bottom的blob为15.申明
virtual void LayerSetUp(const vector
*>& bottom,const vector *>& top);
virtual void Reshape(const vector
*>& bottom,const vector *>& top);
virtual void Forward_cpu(const vector
*>& bottom,const vector *>& top);
virtual void Backward_cpu(const vector
*>& top,const vector & propagate_down, const vector *>& bottom); 6.要是需要GPU加速,则需申明:
virtual void Forward_gpu(const vector
*>& bottom,const vector *>& top);
virtual void Backward_gpu(const vector
*>& top,const vector & propagate_down, const vector *>& bottom); 7.其他(根据算法需要的函数以及参数)
可以在/caffe/include/caffe/下找到许多对应的例子,比如inner_product_layer.hpp:
#ifndef CAFFE_INNER_PRODUCT_LAYER_HPP_
#define CAFFE_INNER_PRODUCT_LAYER_HPP_
#include
#include "caffe/blob.hpp"
#include "caffe/layer.hpp"
#include "caffe/proto/caffe.pb.h"
namespace caffe {
/**
* @brief Also known as a "fully-connected" layer, computes an inner product
* with a set of learned weights, and (optionally) adds biases.
*
* TODO(dox): thorough documentation for Forward, Backward, and proto params.
*/
template <typename Dtype>
class InnerProductLayer : public Layer {
public:
explicit InnerProductLayer(const LayerParameter& param)
: Layer(param) {}
virtual void LayerSetUp(const vector *>& bottom,
const vector *>& top);
virtual void Reshape(const vector *>& bottom,
const vector *>& top);
virtual inline const char* type() const { return "InnerProduct"; }
virtual inline int ExactNumBottomBlobs() const { return 1; }
virtual inline int ExactNumTopBlobs() const { return 1; }
protected:
virtual void Forward_cpu(const vector *>& bottom,
const vector *>& top);
virtual void Forward_gpu(const vector *>& bottom,
const vector *>& top);
virtual void Backward_cpu(const vector *>& top,
const vector<bool>& propagate_down, const vector *>& bottom);
virtual void Backward_gpu(const vector *>& top,
const vector<bool>& propagate_down, const vector *>& bottom);
int M_;
int K_;
int N_;
bool bias_term_;
Blob bias_multiplier_;
bool transpose_; ///< if true, assume transposed weights
};
} // namespace caffe
#endif // CAFFE_INNER_PRODUCT_LAYER_HPP_
1 添加你的源文件置于 src/caffe/layers/下,比如 src/caffe/layers/your_layer.cpp
2.实现LayerSetUp方法(在这里你可以读取layer的参数,权重进行初始化等等),该方法在layer::SetUp时候被调用,用于layer的初始化
3.实现Reshape 方法,根据bottom的shape,修改top的shape等等,也是在layer::SetUp时候被调用,用于layer的初始化
4.实现Forward_cpu和Backward_cpu 方法,前向传播计算loss和top,反向传播计算diff(梯度)
5.在文件末尾加上这两行代码(XXXLayer表示layer的类名),以此在fayer_factory.hpp中注册了此layer以便于运行时的统一创建
INSTANTIATE_CLASS(XXXLayer);
REGISTER_LAYER_CLASS(XXX);
具体例子可参考here
1.如果需要gpu加速的话,那么你需要在src/caffe/layers/下创建.cu文件,比如src/caffe/layers/your_layer.cu
2.采用cuda语言编程,实现Forward_gpu和Backward_gpu方法,和.cpp文件中Forward_cpu和Backward_cpu 方法实现类似,需要把所有的cpu字样改成gpu
具体例子参考here
1.如果想要在net.prototxt中设置你的layer的参数的话,你需要在caffe.proto中定义,定义好之后,即可在forward或者backward的方法中获取到参数值,进行其他相关运算
一个简单的例子(InnerProductLayer )如下:
message InnerProductParameter {
optional uint32 num_output = 1; // The number of outputs for the layer
optional bool bias_term = 2 [default = true]; // whether to have bias terms
optional FillerParameter weight_filler = 3; // The filler for the weight
optional FillerParameter bias_filler = 4; // The filler for the bias
// The first axis to be lumped into a single inner product computation;
// all preceding axes are retained in the output.
// May be negative to index from the end (e.g., -1 for the last axis).
optional int32 axis = 5 [default = 1];
// Specify whether to transpose the weight matrix or not.
// If transpose == true, any operations will be performed on the transpose
// of the weight matrix. The weight matrix itself is not going to be transposed
// but rather the transfer flag of operations will be toggled accordingly.
optional bool transpose = 6 [default = false];
}
2.与此同时,在caffe.proto 的message LayerParameter中添加对应的消息,同时更新一下注释,表明下一个可用的数字大小,比如:
// LayerParameter next available layer-specific ID: 117 (last added: inner_product_param )
message LayerParameter {
...
...
...
optional InnerProductParameter inner_product_param = 117;
...
...
...
}
最后重新编译一下caffe代码即可
CAFFE_ROOT$
make clean
make all