training problem

[GenjiCy]

I would like to use this model for multiple-camera people-trakcing, but i see none training-guide. Is this could just use yolo-format labels to train?

[Ayushpanchratan]

How to train this model on custom dataset ??

[ani8897]

You can use any custom object detector and tracker for detecting and tracking people with bounding boxes in a single camera view.

For multi-camera tracking, this method uses the homography between the two camera views for associating tracking IDs across views. Please refer to the Homography Estimation section in https://arxiv.org/pdf/2111.10971.pdf for more details.