Convert pytorch to Caffe by ONNX

This tool converts pytorch model to Caffe model by ONNX
only use for inference

Dependencies

• caffe (with python support)

gpu: conda install -c defaults caffe-gpu
cpu: conda install -c defaults caffe

• pytorch 0.4 (optional if you only want to convert onnx)
• onnx

we recomand using protobuf 2.6.1 and install onnx from source

git clone --recursive https://github.com/onnx/onnx.git
cd onnx 
python setup.py install

How to use

run test.py to make sure it has been installed correctly
To convert onnx model to caffe:

git clone https://github.com/owphoo/onnx2caffe.git --depth 1

# test 
bash ./download_sample_model.sh
python convertCaffe.py ./model/MobileNetV2.onnx ./model/MobileNetV2.prototxt ./model/MobileNetV2.caffemodel

Current support operation

• Conv
• ConvTranspose
• BatchNormalization
• MaxPool
• AveragePool
• Relu
• Sigmoid
• Dropout
• Gemm (InnerProduct only)
• Add
• Mul
• Reshape
• Upsample
• Concat
• Flatten
• Slice