MaskRCNN-Pytorch v1.0.0

Introduction:

Serves as the release version 1.0 of our repository, the initial release includes the following features: training MaskRCNN on the PyTorch platform, exporting the model to ONNX format, and deploying the model using C++. The C++ deployment relies on ONNX Runtime for model inference and supports both CPU and GPU execution. The current version of ONNX Runtime used in this release is v-1.12. Precompiled libraries of ONNX Runtime for different platforms can be found at the following links:

1. ONNX Runtime 1.12 for CPU:
   onnxruntime-linux-aarch64-1.12.1.zip
   onnxruntime-linux-x64-1.12.1.zip
   onnxruntime-win-x64-1.12.1.zip

2. ONNX Runtime 1.12 for GPU:
   onnxruntime-linux-aarch64-gpu-1.12.1.zip
   onnxruntime-linux-x64-gpu-1.12.1.zip
   onnxruntime-win-x64-gpu-1.12.1.zip

Note: Current onnxruntime-linux-aarch64-gpu-1.12.1.zip only support on platform NVIDIA Jetson AGX Orin.

Please note that it is essential to choose the appropriate library based on your target platform and hardware configuration. Download the corresponding dependency package and place it in the deploy/onnx_cpp/third-party folder.