An On-Line and Adaptive Method for Detecting Abnormal Events in Videos Using Spatio-Temporal ConvNet

Samir Bouindour,Tian Wang,Mohamad Hittawe,Hichem Snoussi,Nacef Tazi

doi:10.3390/app9040757

Abstract

We address in this paper the problem of abnormal event detection in video-surveillance. In this context, we use only normal events as training samples. We propose to use a modified version of pretrained 3D residual convolutional network to extract spatio-temporal features, and we develop a robust classifier based on the selection of vectors of interest. It is able to learn the normal behavior model and detect potentially dangerous abnormal events. This unsupervised method prevents the marginalization of normal events that occur rarely during the training phase since it minimizes redundancy information, and adapt to the appearance of new normal events that occur during the testing phase. Experimental results on challenging datasets show the superiority of the proposed method compared to the state of the art in both frame-level and pixel-level in anomaly detection task.

Highlights

Abnormal event detection and localization is a challenging and exciting task in video monitoring.the security context in recent years has led to the proliferation of surveillance cameras, which generate large amounts of data
Our combination of pretrained 3D-convolutional neural networks (CNNs) and the proposed classifier is an improvement of our previous work [18] where we introduced a method for abnormalities detection based on a 2D-CNN
We propose in this paper a 3D-CNN combined with a new classifier to overcome these limits: From one side, the 3D-CNN is more suitable for spatiotemporal abnormal event detection

Summary

Introduction

Abnormal event detection and localization is a challenging and exciting task in video monitoring. The security context in recent years has led to the proliferation of surveillance cameras, which generate large amounts of data. This flow of CCTV images creates a lack of efficiency of human operators. After only 20 min of focus, the attention of most human operators decreases to well below acceptable levels [2]. This can lead to potential security breaches, especially when monitoring crowded scene videos

Methods

Results

Conclusion