VideoLights: Feature Refinement and Cross-Task Alignment Transformer for Joint Video Highlight Detection and Moment Retrieval

Benchmark Model Rank Results
highlight-detection-on-qvhighlightsVideoLights-B-pt#4mAP: 42.84Hit@1: 70.56
moment-retrieval-on-charades-staVideoLights-B-pt#10R@1 IoU=0.5: 61.96R@1 IoU=0.7: 41.05R@1 IoU=0.3: 73.33mIoU: 52.94
moment-retrieval-on-qvhighlightsVideoLights-B-pt#7mAP: 47.94R@1 IoU=0.5: 70.36R@1 IoU=0.7: 55.25mAP@0.5: 69.53