search

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
334 stars 35 forks source link

Apperance Model performance very bad in night scenario ? #40

Open CarlHuangNuc opened 1 year ago

CarlHuangNuc commented 1 year ago

Hi Author,

         I found Apperance Model can work in many scenario. but in night ,the picture quality or light have big impact on Apperance model peformance . Why it is normal ？？ because  most of  image net data  came from daytime scene ?
Zhongdao commented 1 year ago

Hi I think you're right. Most of the provided appearance models are trained on ImageNet, it is very likely that images/videos in your scenario are out of the training distribution.