extract features for U-Net using C++, modified codes based on extract_features.cpp
-
read a volume of 3d raw data
-
put the data array into memory_data_layer add a new function AddRawData in memory_data_layer.cpp
template void MemoryDataLayer::AddRawData(float* mat_vector, const vector& labels, int num) { //size_t num = mat_vector.size(); //size_t num = batch_size_; std::cout << "nSlice：" << num << endl; std::cout << "nchannels：" << channels_ << endl; std::cout << "nheight：" << height_ << endl; std::cout << "nwidth：" << width_ << endl;
CHECK(!has_new_data_) <<
"Can't add mat until current data has been consumed.";
CHECK_GT(num, 0) << "There is no mat to add";
CHECK_EQ(num % batch_size_, 0) <<
"The added data must be a multiple of the batch size.";
added_data_.Reshape(num, channels_, height_, width_);
added_label_.Reshape(num, 1, 1, 1);
// Apply data transformations (mirror, scale, crop...)
//this->data_transformer_->Transform(mat_vector, &added_data_);// this layer doesn't need transform
//Dtype* transformed_data = transformed_blob->mutable_cpu_data();
Dtype* feature_blob_data = added_data_.mutable_cpu_data();
for (int i = 0; i < num*channels_*height_*width_; i++){
feature_blob_data[i] = mat_vector[i];
}
//// save images for checking
//FILE * fpFeatures = NULL;
////fopen_s(&fpFeatures, "D:\\iyuzu\\UNet\\unet-master\\image.raw", "wb");
//fopen_s(&fpFeatures, "feature_blob.raw", "wb+");//93-A2-PDW-mag2_new
//if (fpFeatures)
//{
// fwrite(feature_blob_data, sizeof(Dtype), 128 * 128 * 60, fpFeatures);
// fclose(fpFeatures);
//}// save images for checking
Dtype* top_label = added_label_.mutable_cpu_data();
for (int item_id = 0; item_id < num; ++item_id) {
//top_label[item_id] = labels[item_id];
top_label[item_id] = labels[0];
}
// num_images == batch_size_
Dtype* top_data = added_data_.mutable_cpu_data();
Reset(top_data, top_label, num);
has_new_data_ = true;
} add declaration in memory_data_layer.hpp
virtual void AddRawData(float* mat_vector, const vector& labels, int num);
re-build caffe
3.re-construct the slices of output feature map into 3d raw data
4.compile the extract_features.cpp and invoke the exe
"This program takes in a trained network and an input data layer, and then"
" extract features of the input data produced by the net.\n"
"Usage: extract_features input_image_names pretrained_net_param"
" feature_extraction_proto_file extract_feature_blob_name1[,name2,...]"
" save_feature_dataset_name1[,name2,...] num_mini_batches input image path"
" [CPU/GPU] [DEVICE_ID=0]\n"
"Note: you can extract multiple features in one pass by specifying"
" multiple feature blob names and dataset names separated by ','."
" The names cannot contain white space characters and the number of blobs"
" and datasets must be equal.";
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent