Open EddieEduardo opened 2 years ago
Hi, when directly training TransTrack on MOT without pretrained model (crowdhuman_final.pth) for 30 epochs, seemes it still cannot correctly localize and boundingbox any objects in the frame.
Is it because the traning iteration is not big enough, or some other reasons ? Thanks for replying.
HI, could you release some instructions on training on coco dataset? Thanks a lot!