Human-Scene Network: A novel baseline with self-rectifying loss for weakly supervised video anomaly detection.
Snehashis MajhiRui DaiQuan KongLorenzo GarattoniGianpiero FrancescaFrançois BrémondPublished in: Comput. Vis. Image Underst. (2024)
Keyphrases
- anomaly detection
- weakly supervised
- network traffic
- network anomaly detection
- video sequences
- detect anomalies
- intrusion detection
- object detectors
- object class
- intrusion detection system
- normal behavior
- topic models
- video frames
- relation extraction
- multiple images
- video data
- unsupervised learning
- semi supervised
- object recognition
- named entities
- moving objects
- support vector