Jetson Nano System Configuration:

• Ubuntu 18.04.5 LTS
• Jetpack 4.5.1

Yolov5 with PyTorch

Install Docker Engine and Docker Compose:

1. Docker Engine - https://docs.docker.com/engine/install/ubuntu/
2. Docker Compose - https://docs.docker.com/compose/install/

Build docker image:

sudo python3 docker_env.py --build

Run docker container:

sudo python3 docker_env.py --run

Run inferences:

1. cd src
2. python3 yolov5_jetson.py It will download the model the first time you run the code

Yolov5 with TensorRT

Generate .wts from pytorch with .pt, or download .wts from model zoo

1. git clone https://github.com/ultralytics/yolov5.git
2. git clone https://github.com/wang-xinyu/tensorrtx.git
3. // download https://github.com/ultralytics/yolov5/releases/download/v5.0/yolov5s.pt
4. cp tensorrtx/yolov5/gen_wts.py yolov5
5. cd yolov5
6. python gen_wts.py -w yolov5s.pt -o yolov5s.wts // a file 'yolov5s.wts' will be generated.

Build tensorrtx/yolov5 and run

1. cd tensorrtx/yolov5/
2. mkdir build
3. cd build
4. cmake ..
5. make
6. sudo ./yolov5 -s yolov5s.wts yolov5s.engine s
7. sudo ./yolov5 -d yolov5s.engine ../samples

General TensorRT instructions:

https://github.com/wang-xinyu/tensorrtx/tree/master/yolov5