How to run a training once the docker image is built?

[IonaGuyomarch]

Hey there,

First of all, thanks you so much for making this yolov10 support for Jetsons. Now here is my problem: I'm new with Docker, and even if I've achieved all installation steps, I don't get how I can run a prediction using the docker image and without using the local URL (http:https://127.0.0.1:7860). I've tried to run this command:
docker run -it --gpus all -v /path/to/my/models:/data yolov10-jetson
But it just starts the local URL.

I've already trained some models using my Windows machine but I would like to be able to do some predictions using my Jetson Xavier and my pre-trained models.

Any advices?

[yuyoujiang]

Hi IonaGuyomarch,

Thank you very much for your interest in Jetson and YOLOv10.
The jetson-examples provide a convenient way to quickly experience YOLOv10 on Jetson.
If you want to use your own model, I think there are two implementation options:

1. Compile a new docker container. The prerequisite is that you need to spend some time learning about Docker. Here is the configuration file I used when compiling the container.
   https://github.com/Seeed-Projects/jetson-examples/blob/main/reComputer/scripts/yolov10/Dockerfile
2. Follow these steps to install the necessary dependencies on Jetson:
   https://github.com/THU-MIG/yolov10?tab=readme-ov-file#installation
   Since the Jetson ARM64 system, some dependencies might need to be recompiled before they can be installed.

[IonaGuyomarch]

Hi,
Thanks a lot for these information! I'm going to learn more about docker then, I think that's what I need.