open_vision_capsules
open_vision_capsules copied to clipboard
Add a tensorrt backend
Add BaseTensorRTBackend that provide support for TensorRT models
I'll add the rest of my comments once there's a clear story for mutli-GPU support.
@apockill I resolved most of your comments but the post-process stuff. The code is tested to be working and the performance is almost the same. I'll continue to work on the GridParser stuff. In the meantime, you can take another look.