Video Anomaly Detection
Video anomaly detection (VAD) aims to automatically identify unusual events in video footage, a crucial task for security, surveillance, and autonomous driving. Current research emphasizes developing robust methods that generalize well across different datasets and scenarios, focusing on techniques like autoencoders, transformers, and graph neural networks, often incorporating multimodal data (RGB, optical flow, audio) and leveraging pre-trained large language and vision models for improved accuracy and explainability. The field's impact stems from its potential to enhance safety and security in various applications by automating the detection of anomalous activities that might otherwise go unnoticed.
Papers
Evaluating the Effectiveness of Video Anomaly Detection in the Wild: Online Learning and Inference for Real-world Deployment
Shanle Yao, Ghazal Alinezhad Noghre, Armin Danesh Pazho, Hamed Tabkhi
An Exploratory Study on Human-Centric Video Anomaly Detection through Variational Autoencoders and Trajectory Prediction
Ghazal Alinezhad Noghre, Armin Danesh Pazho, Hamed Tabkhi
Harnessing Large Language Models for Training-free Video Anomaly Detection
Luca Zanella, Willi Menapace, Massimiliano Mancini, Yiming Wang, Elisa Ricci
Collaborative Learning of Anomalies with Privacy (CLAP) for Unsupervised Video Anomaly Detection: A New Baseline
Anas Al-lahham, Muhammad Zaigham Zaheer, Nurbek Tastan, Karthik Nandakumar