mobileNet2tensorRT convert Pytorch mobileNet to tensorRt if you want to use c++ code,please follow: cd build cmake .. make then you can run compiled program which can generate .onnx Then you could run mobileNetTrt.py which can use tensorRt inference.