FlashVTG: Feature Layering and Adaptive Score Handling Network for Video Temporal Grounding

Benchmark Model Rank Results
highlight-detection-on-qvhighlightsFlashVTG#2mAP: 44.09Hit@1: 71.01
highlight-detection-on-tvsumFlashVTG#1mAP: 88
highlight-detection-on-youtube-highlightsFlashVTG#5mAP: 75.4
moment-retrieval-on-charades-staFlashVTG#3R@1 IoU=0.5: 70.32R@1 IoU=0.7: 49.87
moment-retrieval-on-qvhighlightsFlashVTG#4mAP: 52.00R@1 IoU=0.5: 70.69R@1 IoU=0.7: 53.96mAP@0.5: 72.33
natural-language-moment-retrieval-on-tacosFlashVTG#6R@1,IoU=0.3: 53.71R@1,IoU=0.5: 41.76R@1,IoU=0.7: 24.74mIoU: 37.61