Sparse-Dense Side-Tuner for efficient Video Temporal Grounding

Benchmark Model Rank Results
moment-retrieval-on-qvhighlightsSDST#3mAP: 53.31mAP@0.5: 71.31mAP@0.75: 54.99
highlight-detection-on-qvhighlightsSDST#4mAP: 43.40Hit@1: 69.13
moment-retrieval-on-charades-staSDST (InternVideo2)#5R@1 IoU=0.5: 72.0R@1 IoU=0.7: 52.6mIoU: 61.2