T2SGrid: Temporal-to-Spatial Gridification for Video Temporal Grounding

Benchmark Model Rank Results
moment-retrieval-on-charades-staQwen2-VL-7B + T2SGrid-FT (extra data)–R@1 IoU=0.5: 60.6R@1 IoU=0.7: 35.9R@1 IoU=0.3: 76.9mIoU: 53.2
temporal-sentence-grounding-on-activitynet-captions-temporal-localization-val-2Qwen2-VL-7B + T2SGrid-FT (extra data)–mIoU: 46.7R1@0.5: 48.4R1@0.7: 29.5