AsyNCE: Disentangling False-Positives for Weakly-Supervised Video Grounding.
Cheng DaYanhao ZhangYun ZhengPan PanYinghui XuChunhong PanPublished in: ACM Multimedia (2021)
Keyphrases
- false positives
- weakly supervised
- weakly labeled
- false negative
- false positive rate
- detection rate
- number of false positives
- topic models
- superpixels
- relation extraction
- semi supervised
- object detectors
- video data
- object class
- video frames
- video sequences
- named entities
- training data
- key frames
- object classes
- information retrieval
- text mining
- object detection
- image retrieval
- pairwise
- feature space