ViLLa: Video Reasoning Segmentation with Large Language Model

Benchmark Model Rank Results
referring-video-object-segmentation-on-ref-davis17ViLLa (InternVideo2-6B) (extra data)#12J&F: 74.3
referring-video-object-segmentation-on-mevisViLLa (InternVideo2-6B) (extra data)#18J&F: 49.4
referring-video-object-segmentation-on-revosViLLa (InternVideo2-6B) (extra data)#18J&F: 57.0
referring-expression-segmentation-on-refer-youtube-vos-2021-public-validationViLLa#26J&F: 66.5J: 64.6F: 68.6