the reason why imagenet_det is needed.

[ghost]

hello.
I'm very thank you for your nice work.

I wonder why image-net detection dataset is needed when I want to train video object detection models such as FGFA.

I prepared Image-net VID dataset but couldn't train my custom model properly since I don't have "imagenet_det" file.
Is there any way to train my custom model without a detection dataset?

In my opinion, the vid dataset already has bounding boxes for objects, so the object detection dataset is not nessesary.

thank you for reading and hope this Issue can help others.