Research map / Video / Video segmentation and tracking
A Background-Agnostic Framework with Adversarial Training for Abnormal Event Detection in Video
Mariana-Iuliana Georgescu, Radu Tudor Ionescu, Fahad Shahbaz Khan, Marius Popescu, Mubarak A. Shah
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add adversarial training with pseudo-abnormal examples and segmentation (this work vs earlier version) | background-agnostic framework · Avenue | micro AUC (%): about 5% higher micro AUC than the preliminary method | Improved |
| this work vs preliminary version | framework · ShanghaiTech | micro AUC (%): 4% margin micro, 4.4% macro | Improved |
| this work vs preliminary version | framework · ShanghaiTech | RBDC (%): more than 20% | Improved |
| train on ShanghaiTech instead of Avenue (cross-domain) | framework · Avenue | micro AUC (%): 92.3 -> 83.6 | Hurt |
| train on Avenue instead of ShanghaiTech (cross-domain) | framework · ShanghaiTech | TBDC (%): 78.79 -> 63.89 | Hurt |
| train on Avenue instead of UCSD Ped2 | framework · UCSD Ped2 | macro AUC (%): 99.7 -> 97.2 | Hurt |
| train on Avenue instead of Subway Exit | framework · Subway Exit | TBDC (%): 67.96 -> 63.93 | Hurt |
| combine Avenue and UCSD Ped2 as source data sets | framework · Subway Exit | frame AUC new labels (%): 92.4 -> 92.3 | No clear effect |
| use new frame labels instead of old labels | framework · Subway Entrance | frame AUC (%): 87.6 -> 92.2 | Improved |
| use reconstructions instead of absolute differences as classifier input (ablation) | binary classifiers · preliminary experiments | accuracy | Hurt |
| use pseudo-abnormal adversarial decoder (adversarial training) (ablation) | auto-encoders · Avenue test video 03 | frame AUC (%): 93.25 -> 98.83 | Improved |
| this work vs Ionescu et al. on Subway Exit (new labels) | framework · Subway Exit | frame AUC (%): 0.9% higher | Improved |
Closest papers by meaning
- Object-Centric Auto-Encoders and Dummy Anomalies for Abnormal Event Detection in Video (CVPR 2019 · similarity 0.94)
- Self-Distilled Masked Auto-Encoders are Efficient Video Anomaly Detectors (CVPR 2024 · similarity 0.88)
- Track Any Anomalous Object:A Granular Video Anomaly Detection Pipeline (CVPR 2025 · similarity 0.87)
- FrameShield: Adversarially Robust Video Anomaly Detection (NeurIPS 2025 · similarity 0.87)
- Prompt-Enhanced Multiple Instance Learning for Weakly Supervised Video Anomaly Detection (CVPR 2024 · similarity 0.87)
- Open-Vocabulary Video Anomaly Detection (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.
