SPOT: Spatiotemporal Prompt Optimization for Motion-Stabilized MLLM-Guided Video Segmentation

Benchmark Model Rank Results
referring-expression-segmentation-on-refer-youtube-vos-2021-public-validationSPOT-13B–J&F: 71.8
referring-video-object-segmentation-on-mevisSPOT-13B–J&F: 51.2
referring-video-object-segmentation-on-revosSPOT-13B–J&F: 54.8