Darknet2Caffemodel

Requirements and Caffe Installation Guidance

Python2.7

$ conda create -n darknet2caffe python=2.7 # Create virutal env has python version of 2.7
$ conda activate darknet2caffe # Activate virtual env
$ conda install cython scikit-image ipython h5py nose pandas protobuf pyyaml # Install Deps
$ cd $HOME/caffe/python
$ for req in $(cat requirements.txt); do pip install $req; done  # Install Requirements

Caffe

Modify Makefile.config and make sure the CONDA_VENV_HOME is set to your anaconda directory
Git Clone

$ cd $HOME && git clone https://github.com/BVLC/caffe.git
$ cd $HOME && git clone https://github.com/es6rc/darknet2caffe.git

Modify caffe source code as Modifications
Make Caffe

$ cp $HOME/darknet2caffe/Makefile.config $HOME/caffe # Make sure $CONDA_VENV_HOME in Makefile.config is your venv directory
$ cd $HOME/caffe
$ make all
$ make pycaffe
$ echo 'export PYTHONPATH=$HOME/caffe/python:$PYTHONPATH' >> $HOME/.bashrc # Add Python Path for Caffe

Pytorch >= 0.40

$ conda activate darknet2caffe # Activate virtual env
$ pip install torch
$ pip install future # Install dependencies

Modifications

Add Caffe Layers for yolov3 and yolov4

Copy caffe_layers/mish_layer/mish_layer.hpp,caffe_layers/upsample_layer/upsample_layer.hpp into include/caffe/layers/.

$ cd $HOME/darknet2caffe
$ cp ./caffe_layers/mish_layer/mish_layer.hpp $HOME/caffe/include/caffe/layers/
$ cp ./caffe_layers/upsample_layer/upsample_layer.hpp $HOME/caffe/include/caffe/layers/

Copy caffe_layers/mish_layer/mish_layer.cpp mish_layer.cu,caffe_layers/upsample_layer/upsample_layer.cpp upsample_layer.cu into src/caffe/layers/.

$ cd $HOME/darknet2caffe
$ cp caffe_layers/mish_layer/mish_layer.c* $HOME/caffe/src/caffe/layers/
$ cp caffe_layers/upsample_layer/upsample_layer.c* $HOME/caffe/src/caffe/layers/

Copy caffe_layers/pooling_layer/pooling_layer.cpp into src/caffe/layers/.Note:only work for yolov3-tiny,use with caution.

$ cd $HOME/darknet2caffe
$ cp caffe_layers/pooling_layer/pooling_layer.cpp $HOME/caffe/src/caffe/layers/

Add below code into src/caffe/proto/caffe.proto.

// LayerParameter next available layer-specific ID: 147 (last added: recurrent_param)
message LayerParameter {
  optional TileParameter tile_param = 138;
  optional VideoDataParameter video_data_param = 207;
  optional WindowDataParameter window_data_param = 129;
++optional UpsampleParameter upsample_param = 149; //added by chen for Yolov3, make sure this id 149 not the same as before.
++optional MishParameter mish_param = 150; //added by chen for yolov4,make sure this id 150 not the same as before.
}
++message VideoDataParameter{
++  enum VideoType {
++    WEBCAM = 0;
++    VIDEO = 1;
++  }
++  optional VideoType video_type = 1 [default = WEBCAM];
++  optional int32 device_id = 2 [default = 0];
++  optional string video_file = 3;
++  // Number of frames to be skipped before processing a frame.
++  optional uint32 skip_frames = 4 [default = 0];
++}
// added by chen for YoloV3
++message UpsampleParameter{
++  optional int32 scale = 1 [default = 1];
++}

// Message that stores parameters used by MishLayer
++message MishParameter {
++  enum Engine {
++    DEFAULT = 0;
++    CAFFE = 1;
++    CUDNN = 2;
++  }
++  optional Engine engine = 2 [default = DEFAULT];
++}

Modify Makefile

Modify Makefile and add -std=c++11 to CXXFLAGS, NVCCFLAGS and LINKFLAGS

### Line 423
CXXFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS) -std=c++11
NVCCFLAGS += -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS) -std=c++11
### Line 426&427
LINKFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS) -std=c++11
### Line 429

6.(re)make caffe.

Demo

  $ python darknet2caffe.py cfg[in] weights[in] prototxt[out] caffemodel[out]

Example

$ mkdir caffemodel # o/w throw core dump error
$ python darknet2caffe.py ./cfg/ods_yolov3.cfg ./weights/ods_yolov3_final.weights ./prototxt/ods_yolov3.prototxt ./caffemodel/ods_yolov3.caffemodel

partial log as below.

I0522 10:19:19.015708 25251 net.cpp:228] layer1-act does not need backward computation.
I0522 10:19:19.015712 25251 net.cpp:228] layer1-scale does not need backward computation.
I0522 10:19:19.015714 25251 net.cpp:228] layer1-bn does not need backward computation.
I0522 10:19:19.015718 25251 net.cpp:228] layer1-conv does not need backward computation.
I0522 10:19:19.015722 25251 net.cpp:228] input does not need backward computation.
I0522 10:19:19.015725 25251 net.cpp:270] This network produces output layer139-conv
I0522 10:19:19.015731 25251 net.cpp:270] This network produces output layer150-conv
I0522 10:19:19.015736 25251 net.cpp:270] This network produces output layer161-conv
I0522 10:19:19.015911 25251 net.cpp:283] Network initialization done.
unknow layer type yolo 
unknow layer type yolo 
save prototxt to prototxt/yolov4.prototxt
save caffemodel to caffemodel/yolov4.caffemodel