Research map / Video / Video segmentation and tracking
Autoregressive Denoising Score Matching is a Good Video Anomaly Detector
Hanwen Zhang, Congqi Cao, Qinyi Lv, Lingtong Min, Yanning Zhang
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 autoregressive denoising score matching to plain DSM (ablation) | NCST · ShanghaiTech | AUC macro (%): 78.5 -> 84.1 | Improved |
| add scene condition embedding (ablation) | ADSM · ShanghaiTech | AUC macro (%): 84.1 -> 86.3 | Improved |
| add key frame motion weighting (ablation) | ADSM · ShanghaiTech | AUC macro (%): 84.1 -> 87.7 | Improved |
| aggregate PSNR visual difference between denoised and original data (ablation) | ADSM · ShanghaiTech | AUC macro (%): 84.1 -> 89.9 | Improved |
| combine scene, motion and appearance components (ablation) | ADSM · ShanghaiTech | AUC macro (%): 89.9 -> 93.2 | Improved |
| apply ADSM in pretrained VAE latent space instead of original pixel space (ablation) | ADSM · ShanghaiTech | AUC macro (%): 93.2 -> 87.5 | Hurt |
| apply ADSM in latent space on NWPU Campus (ablation) | ADSM · NWPU Campus | AUC macro (%): 78.1 -> 73.2 | Hurt |
| use ADSM instead of other likelihood-based methods (MSMA, MULDE) | Likelihood-based VAD · ShanghaiTech | AUC macro (%): 90.4 -> 93.2 | Improved |
| use ADSM versus skeleton-based STG-NF on micro AUC | VAD methods · ShanghaiTech | AUC micro (%): 85.9 -> 84.5 | Hurt |
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.
- MULDE — MULDE: Multiscale Log-Density Estimation via Denoising Score Matching for Video Anomaly Detection (CVPR 2024) · better on 2 of 2 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
- MULDE: Multiscale Log-Density Estimation via Denoising Score Matching for Video Anomaly Detection (CVPR 2024 · similarity 0.90)
- Track Any Anomalous Object:A Granular Video Anomaly Detection Pipeline (CVPR 2025 · similarity 0.87)
- Video Anomaly Detection with Motion and Appearance Guided Patch Diffusion Model (AAAI 2025 · similarity 0.87)
- Dual Conditioned Motion Diffusion for Pose-Based Video Anomaly Detection (AAAI 2025 · similarity 0.86)
- FrameShield: Adversarially Robust Video Anomaly Detection (NeurIPS 2025 · similarity 0.86)
- Advancing Video Anomaly Detection: A Concise Review and a New Dataset (NeurIPS 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.
