This repository builds docker image for object detection using Yolov5 on Nvidia Jetson platform. All operations below should be done on Jetson platform.

This operation makes default docker runtime 'nvidia'.

./setup.sh

This operation build docker image named 'yolov5'. Make sure this takes a few hours.

./build.sh

real 140m18.467s
user 0m2.160s
sys 0m1.524s

If the build fails, use --no-cache option to clean the docker build cache.

./build.sh --no-cache

This operation detects objects with camera connected to /dev/video0.

./run.sh

You can use your own weights(my-weights.pt), as follows:

mkdir -p /path/to/weights
cp my-weights.pt /path/to/weights
xhost +local:
docker run -it --rm \
           --runtime nvidia \
           --network host \
           --device /dev/video0:/dev/video0:mrw \
           -e DISPLAY=$DISPLAY \
           -v /tmp/.X11-unix/:/tmp/.X11-unix \
           -v /path/to/weights:/weights \
           yolov5 python3 detect.py --source 0 --weights /weights/my-weights.pt

for more details: https://github.com/JetsonHacksNano/CSI-Camera

./run-csi-camera.sh