roboflow / zero-shot-object-tracking

Object tracking implemented with the Roboflow Inference API, DeepSort, and OpenAI CLIP.
https://blog.roboflow.com/zero-shot-object-tracking/
GNU General Public License v3.0
355 stars 61 forks source link

Adding Support for YoloR #6

Closed VYRION-Ai closed 2 years ago

VYRION-Ai commented 2 years ago

How can i apply this with YoloR Model

yeldarby commented 2 years ago

We're working on adding YOLOv4 support next, but PRs welcome to add support for other models!

Here was the PR which added YOLOv5 support that you can use as a template: https://github.com/roboflow-ai/zero-shot-object-tracking/pull/5

Jacobsolawetz commented 2 years ago

Yes @totoadel - the YOLOv5 PR that @yeldarby suggested will be a good template to work from, as most of the YOLOR repo, including the inference routines were forked from YOLOv5.

It would also be nice to know that your PR can implement YOLOR inference with a custom model, trained from a tutorial like this one https://blog.roboflow.com/train-yolor-on-a-custom-dataset/

Best of luck and keep us posted in this thread

VYRION-Ai commented 2 years ago

@Jacobsolawetz @yeldarby Do you prefer YoloR or Yolov5?

Jacobsolawetz commented 2 years ago

@totoadel Despite the COCO evaluations, I have seen superior custom training speed and results from YOLOv5 and recommend you use it.

VYRION-Ai commented 2 years ago

@Jacobsolawetz Thank you so much< i will use yolov5 in my work>