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

HOTA vs HOTA(0) #10

Closed philip-fu closed 2 years ago

philip-fu commented 2 years ago

Hi,

For evaluation, the HOTA printed out and the HOTA(0) in the csv don't match. What does the '(0)' mean?

Thanks,

Zhongdao commented 2 years ago

I think this means the HOTA value at Loc-IoU threshold \alpha=0. The overall HOTA value is an average over HOTA(\alpha) with \alpha ranging from 0 to 1. See their blog. I didn't check this though, and always stick with the printed HOTA.

philip-fu commented 2 years ago

Got it. Thanks Zhongbao