Research map / Detection and segmentation / Open-set and few-shot detection
Anomaly Detection in Video via Self-Supervised and Multi-Task Learning
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Fuse object-level and frame-level anomaly scores (ablation) | Four-task framework · Avenue | frame-level AUC (%): 91.9 -> 92.8 | Helped |
| Use the framework at frame level rather than object level (ablation) | Four-task framework · Avenue | frame-level AUC (%): 91.9 -> 86.9 | Hurt |
| Fuse object-level and frame-level anomaly scores (ablation) | Four-task framework · ShanghaiTech | frame-level AUC (%): 89.3 -> 90.2 | Helped |
| Use the framework at frame level rather than object level (ablation) | Four-task framework · ShanghaiTech | frame-level AUC (%): 89.3 -> 83.5 | Hurt |
| Use the framework at frame level rather than object level (ablation) | Four-task framework · UCSD Ped2 | frame-level AUC (%): 99.8 -> 92.4 | Hurt |
| Add motion irregularity prediction to arrow-of-time learning (ablation) | Object-level framework · Avenue | frame-level AUC (%): 83.6 -> 87.7 | Helped |
| Add middle box prediction to Tasks 1+2 (ablation) | Object-level framework · Avenue | frame-level AUC (%): 87.7 -> 91.6 | Helped |
| Add knowledge distillation to Tasks 1+2+3 (ablation) | Object-level framework · Avenue | frame-level AUC (%): 91.6 -> 91.9 | Helped |
| Add motion irregularity prediction to arrow-of-time learning (ablation) | Object-level framework · UCSD Ped2 | frame-level AUC (%): 89.4 -> 97.2 | Helped |
| Add middle box prediction to Tasks 1+2 (ablation) | Object-level framework · UCSD Ped2 | frame-level AUC (%): 97.2 -> 98.8 | Helped |
| Add knowledge distillation to Tasks 1+2+3 (ablation) | Object-level framework · UCSD Ped2 | frame-level AUC (%): 98.8 -> 99.8 | Helped |
| Use knowledge distillation as the only proxy task (ablation) | Object-level framework · Avenue | frame-level AUC (%): 83.6 -> 73.7 | Hurt |
Closest papers by meaning
- MIST: Multiple Instance Self-Training Framework for Video Anomaly Detection (CVPR 2021 · similarity 0.90)
- Object-Centric Auto-Encoders and Dummy Anomalies for Abnormal Event Detection in Video (CVPR 2019 · similarity 0.89)
- Self-Distilled Masked Auto-Encoders are Efficient Video Anomaly Detectors (CVPR 2024 · similarity 0.87)
- Open-Vocabulary Video Anomaly Detection (CVPR 2024 · similarity 0.87)
- Prompt-Enhanced Multiple Instance Learning for Weakly Supervised Video Anomaly Detection (CVPR 2024 · similarity 0.87)
- Multi-Scale Video Anomaly Detection by Multi-Grained Spatio-Temporal Representation Learning (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.