Closed xxxpsyduck closed 3 years ago
I trying to do inference with video input. After running inference on a frame, do I need to destroy the stream and recreate a new one for new frame?
no need
Here is an TensorRT inference on video example which includes detection and tracking: https://github.com/linghu8812/tensorrt_tracker
I trying to do inference with video input. After running inference on a frame, do I need to destroy the stream and recreate a new one for new frame?