Research map / Video / Video segmentation and tracking
Learning to Tell Apart: Weakly Supervised Video Anomaly Detection via Disentangled Semantic Alignment
Wenti Yin, Huaxin Zhang, Xiang Wang, Yuqing Lu, Yicheng Zhang, Bingquan Gong, Jialong Zuo, Li Yu, Changxin Gao, Nong Sang
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 Adapter to VadCLIP baseline (ablation) | VadCLIP · XD-Violence | AP / average mAP: 84.51 -> 85.00 AP; 24.70 -> 28.15 AVG | Improved |
| add Self-Guided Normality Modeling on top of Adapter (ablation) | VadCLIP + Adapter · XD-Violence | AP / average mAP: 85.00 -> 85.94 AP; 28.15 -> 28.39 AVG | Improved |
| add Decoupled Contrastive Semantic Alignment on top of Adapter (ablation) | VadCLIP + Adapter · XD-Violence | AP / average mAP: 85.00 -> 85.67 AP; 28.15 -> 28.25 AVG | Improved |
| add both SG-NM and DCSA to Adapter model (full DSANet) (ablation) | VadCLIP + Adapter · XD-Violence | AP / average mAP: 85.00 -> 86.95 AP; 28.15 -> 28.87 AVG | Improved |
| use Adapter text tuning instead of frozen text encoder (ablation) | DSANet · XD-Violence | AP: 81.57 -> 86.95 | Improved |
| use manual prompt instead of frozen class labels (ablation) | DSANet text encoder · XD-Violence | AP / average mAP: 81.57 -> 81.05 AP; 27.60 -> 28.05 AVG | Mixed |
| use learnable prompt instead of frozen class labels (ablation) | DSANet text encoder · XD-Violence | AP / average mAP: 81.57 -> 82.88 AP; 27.60 -> 28.26 AVG | Improved |
| DSANet versus best prior method on fine-grained detection | DSANet · UCF-Crime | average mAP: 9.62 -> 13.01 | Improved |
| decoupled background prototype aligned to normal class | DSANet vs VadCLIP · UCF-Crime | background alignment accuracy: 87.63% -> 99.63% | Improved |
Closest papers by meaning
- VadCLIP: Adapting Vision-Language Models for Weakly Supervised Video Anomaly Detection (AAAI 2024 · similarity 0.90)
- Learning Event Completeness for Weakly Supervised Video Anomaly Detection (ICML 2025 · similarity 0.90)
- Prompt-Enhanced Multiple Instance Learning for Weakly Supervised Video Anomaly Detection (CVPR 2024 · similarity 0.89)
- Harnessing Large Language Models for Training-free Video Anomaly Detection (CVPR 2024 · similarity 0.89)
- Ex-VAD: Explainable Fine-grained Video Anomaly Detection Based on Visual-Language Models (ICML 2025 · similarity 0.88)
- Open-Vocabulary Video Anomaly Detection (CVPR 2024 · similarity 0.88)
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.
