VITA: Video Instance Segmentation via Object Token Association

Benchmark Model Rank Results
video-instance-segmentation-on-ovis-validationVITA (Swin-L)#29mask AP: 27.7AP50: 51.9AP75: 24.9AR1: 14.9AR10: 33.0
video-instance-segmentation-on-youtube-vis-2021VITA (Swin-L)#11mask AP: 57.5AP50: 80.6AP75: 61.0AR1: 47.7AR10: 62.6