search

xingyizhou / GTR

Global Tracking Transformers, CVPR 2022
374 stars 57 forks source link

CenterNet vs CenterNet2 #34

Open briannlongzhao opened 2 years ago

briannlongzhao commented 2 years ago

Hi, I see that in the configs, some models are using CenterNet while the others are using CenterNet2, is there any reason for using one over the other in different models? thank you!

xingyizhou commented 2 years ago

Hi,

Models on MOT (single-class tracking) use CenterNet, and models on TAO (large-vocabulary tracking) use CenterNet2. In my experience, two-stage models are easier to work in large-vocabulary settings.

Best, Xingyi

briannlongzhao commented 2 years ago

Thanks for your response! If I want to evaluate on BDD100K dataset, do you have any suggestions on which approach I should choose?