| Video Anomaly Detection: A Comprehensive Survey of Deep Learning Approaches |
| Paper ID : 1079-ICEEM2025 (R1) |
| Authors |
|
Ahmed Azab *1, Mohamed A. El-Rashidy2 1Ahmed Azab, Senior Data Engineer, 2proffessor |
| Abstract |
| Video anomaly detection (VAD) plays a crucial role in intelligent surveillance systems, aiming to identify abnormal events that deviate from usual patterns in real-world environments. With the proliferation of deep learning technologies, significant progress has been made using unsupervised and weakly supervised learning approaches that address the scarcity of labeled anomalous data. This survey provides a comprehensive overview of the recent advancements in VAD, categorizing methods based on reconstruction and prediction paradigms and highlighting key innovations in generative models such as autoencoders, variational autoencoders (VAEs), and generative adversarial networks (GANs). We further explore advanced spatiotemporal learning techniques, attention mechanisms, and multi-stream architectures that capture complex temporal dependencies and spatial relationships. A detailed review of benchmark datasets and evaluation metrics is presented, followed by a comparative analysis of state-of-the-art methods across different scenarios. Finally, we discuss open challenges including real-time processing, cross-domain generalization, and interpretability, along with future research directions to guide continued advancement in this rapidly evolving domain. |
| Keywords |
| Video Anomaly Detection; Generative adversarial network; deep learning. |
| Status: Accepted |