Correlation-Guided Query-Dependency Calibration for Video Temporal Grounding

Benchmark Model Rank Results
highlight-detection-on-tvsumCG-DETR#3mAP: 86.8
highlight-detection-on-youtube-highlightsCG-DETR#4mAP: 75.9
highlight-detection-on-qvhighlightsCG-DETR (w/ PT)#8mAP: 40.71Hit@1: 66.60
highlight-detection-on-qvhighlightsCG-DETR#11mAP: 40.33Hit@1: 66.21
moment-retrieval-on-qvhighlightsCG-DETR (w/ PT)#8mAP: 47.97R@1 IoU=0.5: 68.48R@1 IoU=0.7: 53.11mAP@0.5: 69.40…
moment-retrieval-on-qvhighlightsCG-DETR#26mAP: 42.86R@1 IoU=0.5: 65.43R@1 IoU=0.7: 48.38mAP@0.5: 64.51…
natural-language-moment-retrieval-on-tacosCG-DETR#10R@1,IoU=0.3: 52.23R@1,IoU=0.5: 39.61R@1,IoU=0.7: 22.23mIoU: 36.48
moment-retrieval-on-charades-staCG-DETR#39R@1 IoU=0.5: 58.44R@1 IoU=0.7: 36.34