Research map / Video / Video segmentation and tracking
PRISM: Training-Free Video Anomaly Detection via Intrinsic Statistical Modeling
YUANTONG CHEN, Zhengyan Ding, YanFeng Shang
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 text-side whitening to global-axis scoring (ablation) | PRISM global axis (Qwen) · UCF-Crime | AUC (%): 86.15 -> 87.81 | Improved |
| add text-side whitening to global-axis scoring (ablation) | PRISM global axis (Qwen) · XD-Violence | AP (%): 75.26 -> 75.69 | Improved |
| add text-side whitening to global-axis scoring (ablation) | PRISM global axis (Qwen) · UCF-Crime | SNR: 1.39 -> 1.50 | Improved |
| add scenario-aware multi-axis aggregation to raw scoring (ablation) | PRISM without whitening · UCF-Crime | AUC (%): 86.15 -> 88.47 | Improved |
| add scenario-aware multi-axis aggregation to raw scoring (ablation) | PRISM without whitening · XD-Violence | AP (%): 75.26 -> 77.30 | Improved |
| add whitening on top of scenario-aware multi-axis (ablation) | PRISM (Qwen) · UCF-Crime | AUC (%): 88.47 -> 88.85 | Improved |
| add whitening on top of scenario-aware multi-axis (ablation) | PRISM (Qwen) · XD-Violence | AP (%): 77.30 -> 77.48 | Improved |
| replace instance memory retrieval with global statistics (ablation) | raw features · UCF-Crime | AUC (%): 81.00 -> 86.15 | Improved |
| replace instance memory retrieval with global statistics (ablation) | raw features · XD-Violence | AP (%): 71.43 -> 75.26 | Improved |
| add Flashback scaled penalization to raw memory (ablation) | raw memory retrieval · UCF-Crime | AUC (%): 81.00 -> 82.67 | Improved |
| add Flashback scaled penalization to raw memory (ablation) | raw memory retrieval · XD-Violence | AP (%): 71.43 -> 72.99 | Improved |
| use Perception Encoder instead of Qwen3-VL-Embedding (ablation) | PRISM · XD-Violence | AP (%): 77.48 -> 80.31 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VADTree — VADTree: Explainable Training-Free Video Anomaly Detection via Hierarchical Granularity-Aware Tree (NeurIPS 2025) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Harnessing Large Language Models for Training-free Video Anomaly Detection (CVPR 2024 · similarity 0.88)
- Open-Vocabulary Video Anomaly Detection (CVPR 2024 · similarity 0.87)
- MoniTor: Exploiting Large Language Models with Instruction for Online Video Anomaly Detection (NeurIPS 2025 · similarity 0.87)
- VADTree: Explainable Training-Free Video Anomaly Detection via Hierarchical Granularity-Aware Tree (NeurIPS 2025 · similarity 0.87)
- Ex-VAD: Explainable Fine-grained Video Anomaly Detection Based on Visual-Language Models (ICML 2025 · similarity 0.86)
- VERA: Explainable Video Anomaly Detection via Verbalized Learning of Vision-Language Models (CVPR 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.
