抽取任意层特征---caffe使用MemoryDataLayer从内存中加载数据

最近在搞caffe的应用，因为很多时候我们需要进行服务器来进行特征的抽取，所以我们需要很将单张图片丢入caffe的网络进行一次传递，这样就诞生了一个从内存中如何加载数据进入caffe的需求，这里我直接贴出代码来先：

[cpp] view
plaincopy

#include <boost/make_shared.hpp>

// these need to be included after boost on OS X

#include <string> // NOLINT(build/include_order)

#include <vector> // NOLINT(build/include_order)

#include <fstream> // NOLINT

#include "caffe/caffe.hpp"

#include <opencv.hpp>

static void CheckFile(const std::string& filename) {

std::ifstream f(filename.c_str());

if (!f.good()) {

f.close();

throw std::runtime_error("Could not open file " + filename);

}

f.close();

}

template <typename Dtype>

caffe::Net<Dtype>* Net_Init_Load(

std::string param_file, std::string pretrained_param_file, caffe::Phase phase)

{

CheckFile(param_file);

CheckFile(pretrained_param_file);

caffe::Net<Dtype>* net(new caffe::Net<Dtype>(param_file,phase));

net->CopyTrainedLayersFrom(pretrained_param_file,0);

return net;

}

#define NetF float

int main()

{

cv::Mat src1;

src1 = cv::imread("test.png");

cv::Mat rszimage;

//// The mean file image size is 256x256, need to resize the input image to 256x256

cv::resize(src1, rszimage, cv::Size(227, 227));

std::vector<cv::Mat> dv = { rszimage }; // image is a cv::Mat, as I'm using #1416

std::vector<int> dvl = { 0 };

caffe::Datum data;

caffe::ReadFileToDatum("D:/work/DestImage/crop/CH0005-00-0019/00028.png", &data);

caffe::Net<NetF>* _net = Net_Init_Load<NetF>("deploy_Test.prototxt", "bvlc_alexnet.caffemodel", caffe::TEST);

caffe::MemoryDataLayer<NetF> *m_layer_ = (caffe::MemoryDataLayer<NetF> *)_net->layers()[0].get();

m_layer_->AddMatVector(dv, dvl);

/*float loss = 0.0;

std::vector<caffe::Blob<float>*> results = _net->ForwardPrefilled(&loss);*/

int end_ind = _net->layers().size();

std::vector<caffe::Blob<NetF>*> input_vec;

_net->Forward(input_vec);

boost::shared_ptr<caffe::Blob<NetF>> outPool5 = _net->blob_by_name("pool5");

std::cout << outPool5->shape()[0] << std::endl;

std::cout << outPool5->shape()[1] << std::endl;

std::cout << outPool5->shape()[2] << std::endl;

std::cout << outPool5->shape()[3] << std::endl;

std::cout << outPool5->num() << std::endl;

std::cout << outPool5->channels() << std::endl;

std::cout << outPool5->width() << std::endl;

std::cout << outPool5->height() << std::endl;

std::cout << outPool5->data_at(0, 0, 0, 0) << std::endl;

std::cout << outPool5->data_at(0, 0, 1, 1) << std::endl;

std::cout << outPool5->data_at(0, 95, 5, 5) << std::endl;

const NetF* pstart = outPool5->cpu_data();

std::cout << m_layer_->width() << std::endl;

return 0;

}

然后是配置文件：

[plain] view
plaincopy

name: "CaffeNet"

layers

{

name: "data"

type: MEMORY_DATA

top: "data"

top: "label"

memory_data_param

{

batch_size: 1

channels: 3

height: 227

width: 227

}

transform_param

{

crop_size: 227

mirror: false

#mean_file:"imagenet_mean.binaryproto"

mean_value: 104

mean_value: 117

mean_value: 123

}

}

layers {

name: "`"

type: CONVOLUTION

bottom: "data"

top: "conv1"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

convolution_param {

num_output: 96

kernel_size: 11

stride: 4

}

}

layers {

name: "relu1"

type: RELU

bottom: "conv1"

top: "conv1"

}

layers {

name: "pool1"

type: POOLING

bottom: "conv1"

top: "pool1"

pooling_param {

pool: MAX

kernel_size: 3

stride: 2

}

}

layers {

name: "norm1"

type: LRN

bottom: "pool1"

top: "norm1"

lrn_param {

local_size: 5

alpha: 0.0001

beta: 0.75

}

}

layers {

name: "conv2"

type: CONVOLUTION

bottom: "norm1"

top: "conv2"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

convolution_param {

num_output: 256

pad: 2

kernel_size: 5

group: 2

}

}

layers {

name: "relu2"

type: RELU

bottom: "conv2"

top: "conv2"

}

layers {

name: "pool2"

type: POOLING

bottom: "conv2"

top: "pool2"

pooling_param {

pool: MAX

kernel_size: 3

stride: 2

}

}

layers {

name: "norm2"

type: LRN

bottom: "pool2"

top: "norm2"

lrn_param {

local_size: 5

alpha: 0.0001

beta: 0.75

}

}

layers {

name: "conv3"

type: CONVOLUTION

bottom: "norm2"

top: "conv3"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

convolution_param {

num_output: 384

pad: 1

kernel_size: 3

}

}

layers {

name: "relu3"

type: RELU

bottom: "conv3"

top: "conv3"

}

layers {

name: "conv4"

type: CONVOLUTION

bottom: "conv3"

top: "conv4"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

convolution_param {

num_output: 384

pad: 1

kernel_size: 3

group: 2

}

}

layers {

name: "relu4"

type: RELU

bottom: "conv4"

top: "conv4"

}

layers {

name: "conv5"

type: CONVOLUTION

bottom: "conv4"

top: "conv5"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

convolution_param {

num_output: 256

pad: 1

kernel_size: 3

group: 2

}

}

layers {

name: "relu5"

type: RELU

bottom: "conv5"

top: "conv5"

}

layers {

name: "pool5"

type: POOLING

bottom: "conv5"

top: "pool5"

pooling_param {

pool: MAX

kernel_size: 3

stride: 2

}

}

layers {

name: "fc6"

type: INNER_PRODUCT

bottom: "pool5"

top: "fc6"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

inner_product_param {

num_output: 4096

}

}

layers {

name: "relu6"

type: RELU

bottom: "fc6"

top: "fc6"

}

layers {

name: "drop6"

type: DROPOUT

bottom: "fc6"

top: "fc6"

dropout_param {

dropout_ratio: 0.5

}

}

layers {

name: "fc7"

type: INNER_PRODUCT

bottom: "fc6"

top: "fc7"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

inner_product_param {

num_output: 4096

}

}

layers {

name: "relu7"

type: RELU

bottom: "fc7"

top: "fc7"

}

layers {

name: "drop7"

type: DROPOUT

bottom: "fc7"

top: "fc7"

dropout_param {

dropout_ratio: 0.5

}

}

layers {

name: "fc8"

type: INNER_PRODUCT

bottom: "fc7"

top: "fc8"

blobs_lr: 1

blobs_lr: 2

weight_decay: 1

weight_decay: 0

inner_product_param {

num_output: 1000

}

}

layers

{

name: "prob"

type: SOFTMAX

bottom: "fc8"

top: "prob"

}

layers

{

name: "output"

type: ARGMAX

bottom: "prob"

top: "output"

}

我的模型使用的是alexnet，例子是用来抽取一个图片在pool5那一层的特征。这样大家使用这个例子可以利用caffe的任意模型抽取任意图片的特征。