Object Detection Training Pipeline (ODTP)

[john-]

I created ODTP to streamline the process to create custom YOLO objection detection models from your Label Studio dataset and publicly available COCO 2017 dataset.

It has nothing directly to do with Viseron. However, I have used it (along with ALTS) to improve the YOLO model I use in Viseron.

The pipeline is intended to be pretty general but I am sure there are some specifics of my use case that got left in that will require some customization. It can be be completely self-hosted.

If you are open to using LLMs you can drop these two docs into one and it should have enough information to assist you in getting running and training/improving a model.