Using custom Yolo model and video inference

MasterBin-IIAU / Unicorn

[ECCV'22 Oral] Towards Grand Unification of Object Tracking

MIT License

953 stars 87 forks source link

Using custom Yolo model and video inference #5

Closed bdubbs09 closed 2 years ago

bdubbs09 commented 2 years ago

Is there a way to upload a custom trained model then use it to track and perform inference in a custom video?
Does Unicorn only accept YoloX, or would it be able to accept a generic (ex Yolo5/6/7) trained model?

This may be addressed somewhere in the code, but I have not found it yet. Any help would be amazing!

MasterBin-IIAU commented 2 years ago

@bdubbs09 Hi, Thanks for your attention.

Users can use Unicorn to track objects in a custom video. We will provide an example later.
For now, Unicorn is tightly coupled with the codebase of YOLOX. But it certainly can work with other detectors like YOLOv5-v7.

bdubbs09 commented 2 years ago

Thanks for addressing!

Do you have any suggestions for an integration point for a different Yolo model? Or at least where to start to build that additional functionality? I don't mind making a PR for a feature, but if you'd rather it be done a specific way, no worries.

Thanks again for the help!

MasterBin-IIAU commented 2 years ago

@bdubbs09 Hi, in my opinion, we need to first reimplement YOLOv5, YOLOX, and YOLOV7 with a unified codebase, if we want to seamlessly switch between them. This may require a very large effort and we may not have enough energy to do this for now. But we highly admire and welcome related PR :)