Hey there @mshamrai I wanted to ask you, about the Visdrone dataset, did you use only the DET ones?
I was wondering if you could also convert the MOT ones into valid bounding boxes. Then train on a mixture of DET and MOT ?
Hi. Thanks for reaching out. Interesting point. I will consider this. Thanks again!
@mshamrai welcome. So the weights you shared were trained on the default CoCo weights + Visdrone DET. is that right?
@werkwerk yes, that's right
· Sign up or log in to comment
