Locate Anything in Videos: Rethinking Efficient Generative Spatio-Temporal Video Grounding

Open paper
Benchmark Model Rank Results
referring-video-object-segmentation-on-reasonvosPTD Qwen3-VL-4B + SAM3#3J&F: 64.7
moment-retrieval-on-charades-staPTD Qwen3-VL-4B#39R@1 IoU=0.5: 58.4R@1 IoU=0.7: 30.1R@1 IoU=0.3: 78.9mIoU: 52.4

Rank counts only results with a code link.