VideoGLaMM: A Large Multimodal Model for Pixel-Level Visual Grounding in Videos

Benchmark Model Rank Results
referring-video-object-segmentation-on-ref-davis17VideoGLaMM (Phi-3 Mini 3.8B, RVOS fine-tuned) (extra data)#22J&F: 69.5
referring-expression-segmentation-on-refer-youtube-vos-2021-public-validationVideoGLaMM (Phi-3 Mini 3.8B, RVOS fine-tuned) (extra data)#24J&F: 66.8
referring-video-object-segmentation-on-mevisVideoGLaMM (Phi-3 Mini 3.8B, RVOS fine-tuned) (extra data)#26J&F: 45.15