Research map / Video / Video segmentation and tracking
Clustering Driven Deep Autoencoder for Video Anomaly Detection
Yunpeng Chang, Zhigang Tu, Wei Xie, Junsong Yuan
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 RGB difference motion autoencoder to the spatial autoencoder (ablation) | Spatial autoencoder · Avenue | frame-level AUC: 71.2% -> 81.4% | Improved |
| add appearance autoencoder to the motion autoencoder (ablation) | Motion autoencoder · Avenue | frame-level AUC: 79.9% -> 81.4% | Mixed |
| add variance attention module (ablation) | Motion + appearance autoencoders · Avenue | frame-level AUC: 81.4% -> 82.8% | Improved |
| add deep k-means cluster (ablation) | Motion + appearance autoencoders · Avenue | frame-level AUC: 81.4% -> 83.5% | Improved |
| add deep k-means cluster to model with variance attention (ablation) | Full model · Avenue | frame-level AUC: 82.8% -> 86.0% | Improved |
| add variance attention to model with deep k-means (ablation) | Full model · Avenue | frame-level AUC: 83.5% -> 86.0% | Improved |
| add deep k-means cluster constraint (ablation) | Full model · UCSD Ped2 | frame-level AUC: 94.5% -> 96.5% | Improved |
| increase cluster number from 4 to 32 (ablation) | Full model · UCSD Ped2 | frame-level AUC: 95.6 -> 96.5% | Improved |
| increase cluster number from 32 to 64 (ablation) | Full model · UCSD Ped2 | frame-level AUC: 96.5% -> 96.4% | No clear effect |
| replace FlowNet2-SD optical flow with RGB difference (ablation) | Motion branch · UCSD Ped2 | speed and AUC: 0.071 -> 0.031 seconds per frame | Improved |
| replace FlowNet2-SD optical flow with RGB difference (ablation) | Motion branch · UCSD Ped2 | frame-level AUC: +2.1% | Improved |
| replace TV-L1 optical flow with RGB difference (ablation) | Motion branch · UCSD Ped2 | frame-level AUC: +2.6% | Improved |
Closest papers by meaning
- Object-Centric Auto-Encoders and Dummy Anomalies for Abnormal Event Detection in Video (CVPR 2019 · similarity 0.88)
- Anomaly Detection in Video Sequence With Appearance-Motion Correspondence (ICCV 2019 · similarity 0.87)
- Robust Anomaly Detection in Videos Using Multilevel Representations (AAAI 2019 · similarity 0.86)
- Self-Trained Deep Ordinal Regression for End-to-End Video Anomaly Detection (CVPR 2020 · similarity 0.86)
- Learning Memory-Guided Normality for Anomaly Detection (CVPR 2020 · similarity 0.86)
- Self-Distilled Masked Auto-Encoders are Efficient Video Anomaly Detectors (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.
