reference
jetson nano: python calls the tensorrt accelerated c++ code of yolov5-v6.0, the inference speed is stable at 12fps, and the video memory occupied is within 0.8G
Install pycuda, tensorrt
sudo ln -s /usr/lib/python3.6/dist-packages/tensorrt ~/miniforge3/envs/safeguard/lib/python3.6/site-packages/tensorrt
Execute python yolov5_det_trt.py
Detection speed is slow