Research map / Video / Video segmentation and tracking
O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning
Mei Yuan, Qi Long, Qifeng Wu, Zhenyang Li, Yizhou Zhao, Lei Wang, Yang Liu, Min Xu
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 object state tracking (ablation) | O-VAD · Phys-AD sticky roller | AUC: 0.484 -> 0.811 | Improved |
| add object state tracking (ablation) | O-VAD · Phys-AD rubber band | F1: 0.065 -> 0.667 | Improved |
| add video-level caption (ablation) | O-VAD · Phys-AD sticky roller | AUC: 0.564 -> 0.811 | Improved |
| add video-level caption (ablation) | O-VAD · Phys-AD rubber band | AUC: 0.590 -> 0.721 | Improved |
| add video-level caption (ablation) | O-VAD · Phys-AD screw | F1: 0.733 -> 0.700 | Hurt |
| add video-level caption (ablation) | O-VAD · Phys-AD liquid | AUC: 0.742 -> 0.776 | No clear effect |
| add structured six-step cognitive reasoning instead of generic prompt (ablation) | O-VAD · Phys-AD sticky roller | AUC: 0.720 -> 0.811 | Improved |
| add post visual verification (ablation) | O-VAD · Phys-AD sticky roller | AUC: 0.633 -> 0.811 | Improved |
| use O-VAD instead of GPT-5 direct prompting | O-VAD · Phys-AD video-level | AUROC: 0.502 -> 0.584 | Improved |
| use training-free O-VAD instead of trained S3R | O-VAD · Phys-AD video-level | AUROC: 0.555 -> 0.584 | Improved |
| use O-VAD instead of GPT-5 direct prompting | O-VAD · LiquidAD video-level | AUROC: 0.465 -> 0.692 | Improved |
| use O-VAD instead of GPT-5 direct prompting | O-VAD · IPAD video-level | AUROC: 0.519 -> 0.565 | Improved |
Closest papers by meaning
- Open-Vocabulary Video Anomaly Detection (CVPR 2024 · similarity 0.87)
- AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models (AAAI 2024 · similarity 0.87)
- Supervised Anomaly Detection for Complex Industrial Images (CVPR 2024 · similarity 0.87)
- Track Any Anomalous Object:A Granular Video Anomaly Detection Pipeline (CVPR 2025 · similarity 0.87)
- Towards Visual Discrimination and Reasoning of Real-World Physical Dynamics: Physics-Grounded Anomaly Detection (CVPR 2025 · similarity 0.86)
- Vad-R1: Towards Video Anomaly Reasoning via Perception-to-Cognition Chain-of-Thought (NeurIPS 2025 · 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.
