C++ inference of a pytorch model on cpu
- Using a resnet model trained on imagenet.
pip install -r requirmets.txt
- Donwload LibTorch cpu & Extract it. link
- Clone repository
git clone https://github.com/Sharwon/cpp-torch-cpu.git
- Cmake
mkdir build && cd build
cmake -DCMAKE_PREFIX_PATH=<full path to libtorch folder> ..
make
- Tracing your pytorch model to torch script
python model_trace.py
- mv
.pt
file into models folder
- Test the app
<app> <model> <image> <label>
sample
as n07873807 pizza, pizza pie, score: 1.41523