Zhongdao / UniTrack

[NeurIPS'21] Unified tracking framework with a single appearance model. It supports Single Object Tracking (SOT), Video Object Segmentation (VOS), Multi-Object Tracking (MOT), Multi-Object Tracking and Segmentation (MOTS), Pose Tracking, Video Instance Segmentation (VIS), and class-agnostic MOT (e.g. TAO dataset).
MIT License
338 stars 34 forks source link

Model zoo link doesn't work #9

Closed philip-fu closed 3 years ago

philip-fu commented 3 years ago

Hi, I am trying to use different pretrained models provided by you model zoo but some of the links are empty. And also wondering if you have the im_mean and im_std for each of them.

Thank you,

Zhongdao commented 3 years ago

Hi @philip-fu I have updated the links, now all models can be downloaded. Regarding im_mean and im_std: Most models use an identical setting to ImageNet pre-trained models, i.e.

im_mean = [0.485, 0.456, 0.406]
im_std = [0.229, 0.224, 0.225]

Several models use different settings. For example, in CRW

im_mean = [0.4914, 0.4822, 0.4465]
im_std = [0.2023, 0.1994, 0.2010]

and in UVC/colorization,

im_mean = [0.5, 0.5, 0.5]
im_std = [0.5, 0.5, 0.5]

Also note that UVC/colorization take Lab images as inputs, rather than RGB, so remember to add use_lab: True in the config file (under the common field).

philip-fu commented 3 years ago

Thanks!!