Deep-Anomaly: Fully Convolutional Neural Network for Fast Anomaly Detection in Crowded Scenes

TL;DR

This paper introduces a fast and accurate anomaly detection method in crowded scenes using a fully convolutional neural network that combines supervised and unsupervised learning for effective localization.

Contribution

It proposes a novel FCN-based architecture that efficiently detects and localizes anomalies in videos by leveraging transfer learning and cascaded detection, improving speed and accuracy.

Findings

Outperforms existing methods in accuracy on benchmark datasets

Achieves high speed and efficiency in anomaly detection

Effectively localizes anomalies in crowded scenes

Abstract

The detection of abnormal behaviours in crowded scenes has to deal with many challenges. This paper presents an efficient method for detection and localization of anomalies in videos. Using fully convolutional neural networks (FCNs) and temporal data, a pre-trained supervised FCN is transferred into an unsupervised FCN ensuring the detection of (global) anomalies in scenes. High performance in terms of speed and accuracy is achieved by investigating the cascaded detection as a result of reducing computation complexities. This FCN-based architecture addresses two main tasks, feature representation and cascaded outlier detection. Experimental results on two benchmarks suggest that detection and localization of the proposed method outperforms existing methods in terms of accuracy.