SlotVTG: Object-Centric Adapter for Generalizable Video Temporal Grounding

Benchmark Model Rank Results
moment-retrieval-on-charades-staSlotVTG 7B (extra data)–R@1 IoU=0.5: 67.6R@1 IoU=0.7: 46.7R@1 IoU=0.3: 79.5mIoU: 58.3