SuperClip Pyramid for Video Temporal Grounding

Benchmark Model Rank Results
highlight-detection-on-qvhighlightsSCVTG–mAP: 44.62
highlight-detection-on-tvsumSCVTG–mAP: 89.4
highlight-detection-on-youtube-highlightsSCVTG–mAP: 77.3
moment-retrieval-on-qvhighlightsSCVTG–mAP: 52.73mAP@0.5: 73.78mAP@0.75: 53.70
video-grounding-on-qvhighlightsSCVTG–R@1,IoU=0.7: 55.77R@1,IoU=0.5: 72.70