docker run --name jetson-inference-native -d \
-e TZ=Asia/Shanghai \
-p 5000:5000 \
--runtime nvidia \
hikariai/jetson-inference-native
curl -X POST -H "Content-Type: application/json" \
-d '{"network": "resnet-18", "url": "https://cdn.download.ams.birds.cornell.edu/api/v1/asset/171633971"}' \
http://localhost:5000/inference
docker logs jetson-inference-native --follow