darknet-yolov3转caffemodel

下载https://github.com/ChenYingpeng/caffe-yolov3

1. 修改CMakeLists.txt

主要修改交叉编译器为x86_64编译，修改自己的caffe路径（其中caffe.proto中增加upsample层），修改自己的opencv路径

cmake_minimum_required(VERSION 2.8)
project(sysDetectSpeed)

# setup tensorRT flags
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")	# -std=gnu++11
set(BUILD_DEPS "YES" CACHE BOOL "If YES, will install dependencies into sandbox.  Automatically reset to NO after dependencies are installed.")

# if this is the first time running cmake, perform pre-build dependency install script (or if the user manually triggers re-building the dependencies)
#if( ${BUILD_DEPS} )
#	message("Launching pre-build dependency installer script...")
#
#	execute_process(COMMAND sh ../CMakePreBuild.sh
#				WORKING_DIRECTORY ${PROJECT_BINARY_DIR}
#				RESULT_VARIABLE PREBUILD_SCRIPT_RESULT)

#	set(BUILD_DEPS "NO" CACHE BOOL "If YES, will install dependencies into sandbox.  #Automatically reset to NO after dependencies are installed." FORCE)
#	message("Finished installing dependencies")
#endif()


# Qt is used to load images (installed by ubuntu-desktop)
find_package(Qt4 REQUIRED)
include(${QT_USE_FILE})
add_definitions(${QT_DEFINITIONS})


# setup CUDA
find_package(CUDA)

set(
	CUDA_NVCC_FLAGS
	${CUDA_NVCC_FLAGS}; 
    -O3 
	-gencode arch=compute_53,code=sm_53
	-gencode arch=compute_62,code=sm_62
	-gencode arch=compute_61,code=sm_61
)



# setup project output paths
set(PROJECT_OUTPUT_DIR  ${PROJECT_BINARY_DIR}/${CMAKE_SYSTEM_PROCESSOR})
set(PROJECT_INCLUDE_DIR ${PROJECT_OUTPUT_DIR}/include)

file(MAKE_DIRECTORY ${PROJECT_INCLUDE_DIR})
file(MAKE_DIRECTORY ${PROJECT_OUTPUT_DIR}/bin)

message("-- system arch:  ${CMAKE_SYSTEM_PROCESSOR}")
message("-- output path:  ${PROJECT_OUTPUT_DIR}")

set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${PROJECT_OUTPUT_DIR}/bin)
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${PROJECT_OUTPUT_DIR}/lib)
set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${PROJECT_OUTPUT_DIR}/lib)


# build C/C++ interface
include_directories(${PROJECT_INCLUDE_DIR} ${GIE_PATH}/include)
include_directories(/usr/include/gstreamer-1.0 /usr/lib/x86_64-linux-gnu/gstreamer-1.0/include /usr/include/glib-2.0 /usr/include/libxml2 /usr/lib/x86_64-linux-gnu/glib-2.0/include/ /usr/local/cuda-8.0/extras/CUPTI/include)
include_directories(${PROJECT_INCLUDE_DIR} 
	/home/caffe/include 
	/home/caffe/build/include 
	/usr/local/include  
	/usr/local/cuda/include  
	/usr/include
)


file(GLOB inferenceSources *.cpp *.cu )
file(GLOB inferenceIncludes *.h )

cuda_add_library(sysDetectSpeed SHARED ${inferenceSources})
target_link_libraries(sysDetectSpeed nvcaffe_parser nvinfer Qt4::QtGui GL gstreamer-1.0 gstapp-1.0)	#GLEW	# gstreamer-0.10 gstbase-0.10 gstapp-0.10 
target_link_libraries(sysDetectSpeed 
	/home/caffe/build/lib/libcaffe.so  
	/usr/lib/x86_64-linux-gnu/libglog.so  
	/usr/lib/x86_64-linux-gnu/libgflags.so.2
    	/usr/lib/x86_64-linux-gnu/libboost_system.so  
	/usr/lib/x86_64-linux-gnu/libGLEW.so.1.13  
)


# transfer all headers to the include directory
foreach(include ${inferenceIncludes})
	message("-- Copying ${include}")
	configure_file(${include} ${PROJECT_INCLUDE_DIR} COPYONLY)
endforeach()


# create symbolic link for network data
execute_process( COMMAND "${CMAKE_COMMAND}" "-E" "create_symlink" "${PROJECT_SOURCE_DIR}/data/networks" "${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/networks" )
  
  
# copy image data
file(GLOB imageData ${PROJECT_SOURCE_DIR}/data/images/*)

foreach(image ${imageData})
	message("-- Copying ${image}")
	file(COPY ${image} DESTINATION ${CMAKE_RUNTIME_OUTPUT_DIRECTORY})
	#configure_file(${include} ${CMAKE_RUNTIME_OUTPUT_DIRECTORY} COPYONLY)
endforeach()


add_subdirectory(detectnet)


#setup opencv
FIND_PACKAGE(OpenCV REQUIRED)

#TARGET_LINK_LIBRARIES(sysDetectSpeed ${OpenCV_LIBS})

target_link_libraries(sysDetectSpeed -I/usr/local/include/opencv -I/usr/local/include -L/usr/local/lib
        -lopencv_core -lopencv_highgui -lopencv_imgproc -lopencv_imgcodecs -lopencv_video -lopencv_videoio)

# install
foreach(include ${inferenceIncludes})
    install(FILES "${include}" DESTINATION include/sysDetectSpeed)
endforeach()

# install the shared library
install(TARGETS sysDetectSpeed DESTINATION lib/sysDetectSpeed EXPORT sysDetectSpeedConfig)

# install the cmake project, for importing
install(EXPORT sysDetectSpeedConfig DESTINATION share/sysDetectSpeed/cmake)

此处主要修改处为：

include_directories(${PROJECT_INCLUDE_DIR} 
    /home/caffe/include 
    /home/caffe/build/include #此处应当注意，如果你原始caffe为make编译，则路径应当为/home/caffe/build/src
    /usr/local/include  
    /usr/local/cuda/include  
    /usr/include
)

target_link_libraries(sysDetectSpeed 
    /home/caffe/build/lib/libcaffe.so  
    /usr/lib/x86_64-linux-gnu/libglog.so  
    /usr/lib/x86_64-linux-gnu/libgflags.so.2
        /usr/lib/x86_64-linux-gnu/libboost_system.so  
    /usr/lib/x86_64-linux-gnu/libGLEW.so.1.13  
)

2.编译upsample层 caffe

增加include/caffe/layers/upsample_layer.hpp

#ifndef CAFFE_UPSAMPLE_LAYER_HPP_
#define CAFFE_UPSAMPLE_LAYER_HPP_

#include <vector>

#include "caffe/blob.hpp"
#include "caffe/layer.hpp"
#include "caffe/proto/caffe.pb.h"

namespace caffe {

template <typename Dtype>
class UpsampleLayer : public Layer<Dtype> {
 public:
  explicit UpsampleLayer(const LayerParameter& param)
      : Layer<Dtype>(param) {}
  virtual void LayerSetUp(const vector<Blob<Dtype>*>& bottom,
      const vector<Blob<Dtype>*>& top);
  virtual void Reshape(const vector<Blob<Dtype>*>& bottom,
      const vector<Blob<Dtype>*>& top);

  virtual inline const char* type() const { return "Upsample"; }
  virtual inline int MinBottomBlobs() const { return 1; }
  virtual inline int MaxBottomBlobs() const { return 1; }
  virtual inline int ExactNumTopBlobs() const { return 1; }

  virtual void Forward_cpu(const vector<Blob<Dtype>*>& bottom,
      const vector<Blob<Dtype>*>& top);
  virtual void Forward_gpu(const vector<Blob<Dtype>*>& bottom,
      const vector<Blob<Dtype>*>& top);
  virtual void Backward_cpu(const vector<Blob<Dtype>*>& top,
      const vector<bool>& propagate_down, const vector<Blob<Dtype>*>& bottom);
  virtual void Backward_gpu(const vector<Blob<Dtype>*>& top,
      const vector<bool>& propagate_down, const vector<Blob<Dtype>*>& bottom);

 private:
  int scale_;
};



}  // namespace caffe

#endif  // CAFFE_UPSAMPLE_LAYER_HPP_

增加src/caffe/layers/upsample_layer.cpp

#include <vector>
#include "caffe/layers/upsample_layer.hpp"

namespace caffe {

template <typename Dtype>
void UpsampleLayer<Dtype>::LayerSetUp(
  const vector<Blob<Dtype>*>& bottom, const vector<Blob<Dtype>*>& top) {
  UpsampleParameter upsample_param = this->layer_param_.upsample_param();
  scale_ = upsample_param.scale();
}

template <typename Dtype>
void UpsampleLayer<Dtype>::Reshape(
  const vector<Blob<Dtype>*>& bottom, const vector<Blob<Dtype>*>& top) {
  vector<int> out_shape;
  for (int i = 0; i < bottom[0]->num_axes(); i++) {
    out_shape.push_back(bottom[0]->shape(i));
  }

  out_shape[bottom[0]->num_axes() - 1] *= scale_;
  out_shape[bottom[0]->num_axes() - 2] *= scale_;
  top[0]->Reshape(out_shape);
}

template <typename Dtype>
void UpsampleLayer<Dtype>::Forward_cpu(const vector<Blob<Dtype>*>& bottom,
    const vector<Blob<Dtype>*>& top) {

  int N = top[0]->shape(0);
  int C = top[0]->shape(1);
  int H = top[0]->shape(2);
  int W = top[0]->shape(3);

  const Dtype *input = bottom[0]->cpu_data();
  Dtype *output = top[0]->mutable_cpu_data();
  for (int n = 0; n < N; n++) {
    for (int c = 0; c < C; c++) {
      for (int h = 0; h < H; h++) {
        for (int w = 0; w < W; w++) {
          int nw = w/scale_;
          int nh = h/scale_;
          int out_idx = (((n * C + c) * H) + h) * W + w;
          int in_idx = (((n * C + c) * (H / scale_)) + nh) * (W / scale_) + nw;
          output[out_idx] = input[in_idx];
        }
      }
    }
  }
}

template <typename Dtype>
void UpsampleLayer<Dtype>::Backward_cpu(const vector<Blob<Dtype>*>& top,
    const vector<bool>& propagate_down, const vector<Blob<Dtype>*>& bottom) {
  int N = bottom[0]->shape(0);
  int C = bottom[0]->shape(1);
  int H = bottom[0]->shape(2);
  int W = bottom[0]->shape(3);
  const Dtype *output_grad = top[0]->cpu_diff();
  Dtype* bottom_diff = bottom[0]->mutable_cpu_diff();
  caffe_set(bottom[0]->count(), Dtype(0), bottom_diff);
  for (int n = 0; n < N; n++) {
    for (int c = 0; c < C; c++) {
      for (int h = 0; h < H; h++) {
        for (int w = 0; w < W; w++) {
          for (int i = 0; i < scale_; i++) {
            for (int j = 0; j < scale_; j++) {
              int nw = w * scale_ + i;
              int nh = h * scale_ + j;
              int out_idx = (((n * C + c) * H) + h) * W + w;
              int in_idx = (((n * C + c) * (H * scale_))
                  + nh) * (W * scale_) + nw;
              bottom_diff[out_idx] += output_grad[in_idx];
            }
          }
        }
      }
    }
  }
}

#ifdef CPU_ONLY
STUB_GPU(UpsampleLayer);
#endif

INSTANTIATE_CLASS(UpsampleLayer);
REGISTER_LAYER_CLASS(Upsample);

}  // namespace caffe

增加src/caffe/layers/upsample_layer.cu

  z = ii % d2;
  ii = ii/d2;
  y = ii % d1;
  ii = ii/d1;
  x = ii;
  w = w/scale_factor;
  z = z/scale_factor;
  d2 /= scale_factor;
  d3 /= scale_factor;
  return (((x*d1+y)*d2)+z)*d3+w;
}

__device__ int translate_idx_inv(
    int ii, int d1, int d2, int d3, int scale_factor, int off_x, int off_y) {
  int x, y, z, w;
  w = ii % d3;
  ii = ii/d3;
  z = ii % d2;
  ii = ii/d2;
  y = ii % d1;
  ii = ii/d1;
  x = ii;
  w = w*scale_factor+off_x;
  z = z*scale_factor+off_y;
  d2 *= scale_factor;
  d3 *= scale_factor;
  return (((x*d1+y)*d2)+z)*d3+w;
}

template <typename Dtype>
__global__ void upscale(const Dtype *input, Dtype *output,
        int no_elements, int scale_factor, int d1, int d2, int d3) {
  int ii = threadIdx.x + blockDim.x * blockIdx.x;
  if (ii >= no_elements) return;
  int ipidx = translate_idx(ii, d1, d2, d3, scale_factor);
  output[ii]=input[ipidx];
}

template <typename Dtype>
__global__ void downscale(Dtype *gradInput_data, const Dtype *gradOutput_data,
                          int no_elements, int scale_factor, int d1, int d2,
                          int d3) {
  int ii = threadIdx.x + blockDim.x * blockIdx.x;
  if (ii >= no_elements) return;
  for (int i = 0; i < scale_factor; i++) {
    for (int j = 0; j < scale_factor; j++) {
      int ipidx = translate_idx_inv(ii, d1, d2, d3, scale_factor, i, j);
      gradInput_data[ii] += gradOutput_data[ipidx];
    }
  }
}



template <typename Dtype>
void UpsampleLayer<Dtype>::Forward_gpu(const vector<Blob<Dtype>*>& bottom,
      const vector<Blob<Dtype>*>& top) {
  int d1, d2, d3;

  d1 = top[0]->shape(1);
  d2 = top[0]->shape(2);
  d3 = top[0]->shape(3);

  int no_elements = top[0]->count();

  upscale<Dtype>  // NOLINT_NEXT_LINE(whitespace/operators)
      <<<CAFFE_GET_BLOCKS(no_elements), CAFFE_CUDA_NUM_THREADS>>>(
      bottom[0]->gpu_data(),
      top[0]->mutable_gpu_data(), no_elements, scale_, d1, d2, d3);
}

template <typename Dtype>
void UpsampleLayer<Dtype>::Backward_gpu(const vector<Blob<Dtype>*>& top,
      const vector<bool>& propagate_down, const vector<Blob<Dtype>*>& bottom) {
  int d1, d2, d3;
  Dtype* bottom_diff = bottom[0]->mutable_gpu_diff();
  d1 = bottom[0]->shape(1);
  d2 = bottom[0]->shape(2);
  d3 = bottom[0]->shape(3);
  int no_elements = bottom[0]->count();
  caffe_gpu_set(bottom[0]->count(), Dtype(0), bottom_diff);
  downscale<Dtype>  // NOLINT_NEXT_LINE(whitespace/operators)
      <<<CAFFE_GET_BLOCKS(no_elements), CAFFE_CUDA_NUM_THREADS>>>(
      bottom_diff, top[0]->gpu_diff(), no_elements, scale_, d1, d2, d3);
}

INSTANTIATE_LAYER_GPU_FUNCS(UpsampleLayer);

}  // namespace caffe

增加src/caffe/test/test_upsample_layer.cpp（此步骤可省略）

#include <cmath>
#include <vector>

#include "boost/scoped_ptr.hpp"
#include "gtest/gtest.h"

#include "caffe/blob.hpp"
#include "caffe/filler.hpp"
#include "caffe/layers/upsample_layer.hpp"
#include "caffe/util/io.hpp"

#include "caffe/test/test_caffe_main.hpp"
#include "caffe/test/test_gradient_check_util.hpp"

using boost::scoped_ptr;

namespace caffe {

template <typename TypeParam>
class UpsampleLayerTest : public MultiDeviceTest<TypeParam> {
  typedef typename TypeParam::Dtype Dtype;

 protected:
  UpsampleLayerTest()
      : blob_bottom_data_(new Blob<Dtype>(2, 5, 2, 2)),
        blob_top_data_(new Blob<Dtype>()) {
    // fill the values
    FillerParameter filler_param;
    filler_param.set_std(10);
    GaussianFiller<Dtype> filler(filler_param);
    filler.Fill(this->blob_bottom_data_);
    blob_bottom_vec_.push_back(blob_bottom_data_);
    blob_top_vec_.push_back(blob_top_data_);
  }

  virtual ~UpsampleLayerTest() {
    delete blob_bottom_data_;
    delete blob_top_data_;
  }
  Blob<Dtype>* const blob_bottom_data_;
  Blob<Dtype>* const blob_top_data_;
  vector<Blob<Dtype>*> blob_bottom_vec_;
  vector<Blob<Dtype>*> blob_top_vec_;
};

TYPED_TEST_CASE(UpsampleLayerTest, TestDtypesAndDevices);

TYPED_TEST(UpsampleLayerTest, TestGradient) {
  typedef typename TypeParam::Dtype Dtype;
  LayerParameter layer_param;
  UpsampleParameter* upsample_param =
      layer_param.mutable_upsample_param();
  upsample_param->set_scale(2);
  UpsampleLayer<Dtype> layer(layer_param);
  GradientChecker<Dtype> checker(1e-2, 1e-2, 1701);
  checker.CheckGradientExhaustive(&layer, this->blob_bottom_vec_,
      this->blob_top_vec_, 0);
}

}  // namespace caffe

最后在src/caffe/proto/caffe.proto中增加

message LayerParameter {

           ……

  optional UpsampleParameter upsample_param = 148;
}


message UpsampleParameter {
  optional int32 scale = 1 [default = 1];
}

重新编译caffe，后续转换需添加编译好的caffe路径

3.yolov3转caffemodel

下载https://github.com/marvis/pytorch-caffe-darknet-convert

原工程支持yolov2

首先安装pytorch：pip install torch==0.4.1

然后将ChenYingpeng/caffe-yolov3/model_convert/darknet2caffe_yolov3.py拷贝到pytorch-caffe-darknet-convert下然后在该目录下执行：

python yolov3_darknet2caffe.py yolov3xxx.cfg yolov3xxx.weights xxx.prototxt xxx.caffemodel

4. 在ChenYingpeng/caffe-yolov3下按照README.d中教程，配合修改好的CMAKEList.txt编译，编译后可测试：

caffe-yolov3/detectnet/detectnet.cpp yolov3.prototxt yolov3.caffemodel

若实在无法编译通过（由于博主没有sudo权限，无法安装opencv3），

可参考使用MobileNet-YOLO（https://github.com/eric612/MobileNet-YOLO）中测试脚本测试，其中yolov3.prototxt需要参考其中相应的yolov3.prototxt做修改，然后测试转出的yolov3.caffemodel