Modality-aware Contrastive Instance Learning with Self-Distillation for Weakly-Supervised Audio-Visual Violence Detection.
Jiashuo YuJinyu LiuYing ChengRui FengYuejie ZhangPublished in: ACM Multimedia (2022)
Keyphrases
- weakly supervised
- audio visual
- multi modal
- object detectors
- topic models
- relation extraction
- semi supervised
- image classification
- active learning
- multiscale
- superpixels
- segmentation algorithm
- multimedia
- object class
- feature extraction
- named entities
- question answering
- information retrieval
- image database
- object detection
- object recognition
- keywords