Mitigating Modality and Language-Style Gaps for Zero-Shot Video Moment Retrieval

Benchmark Model Rank Results
moment-retrieval-on-charades-staSelf-SiMS–R@1 IoU=0.5: 39.7R@1 IoU=0.7: 21.0R@1 IoU=0.3: 62.7mIoU: 41.9
moment-retrieval-on-qvhighlightsSelf-SiMS–mAP: 38.3R@1 IoU=0.5: 59.7R@1 IoU=0.7: 42.2mAP@0.5: 59.2