Summary of Da-flow: Dual Attention Normalizing Flow For Skeleton-based Video Anomaly Detection, by Ruituo Wu et al.
DA-Flow: Dual Attention Normalizing Flow for Skeleton-based Video Anomaly Detection
by Ruituo Wu, Yang Chen, Jian Xiao, Bing Li, Jicong Fan, Frédéric Dufaux, Ce Zhu, Yipeng Liu
First submitted to arxiv on: 5 Jun 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed Dual Attention Module (DAM) combines graph convolutional networks (GCN) and temporal convolutional networks (TCN) to capture cross-dimension interaction relationships in spatio-temporal skeletal data, which improves performance in skeleton-based video anomaly detection (SVAD). The DAM employs frame attention mechanisms to identify significant frames and skeleton attention mechanisms to capture broader relationships across fixed partitions with minimal parameters. This module is integrated into the normalizing flow framework as a post-processing unit, resulting in competitive or better performance compared to existing state-of-the-art methods in terms of micro AUC metric. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The researchers combined two types of networks (TCN and GCN) to detect unusual movements in videos that show people’s skeletons. They came up with a new way to use these networks together, which helps them understand how different parts of the video are related. This helps the model find weird movements more accurately. The new method works well even when there is noise or errors in the data. It’s better than other methods at detecting unusual movements and uses fewer computer resources. |
Keywords
» Artificial intelligence » Anomaly detection » Attention » Auc » Gcn
