Efficient Anomaly Detection in Crowd Videos Using Pre-Trained 2D Convolutional Neural Networks

Abid Mehmood

doi:10.1109/access.2021.3118009

Abstract

Surveillance of crowded places can benefit from improved techniques of anomaly detection in crowd videos. Several existing methods have detected various types of crowd abnormal behaviors by using spatial and temporal information got from videos. So far as real-time detection of anomalies is concerned, special attention must be given to reducing the model complexity that leads to computational and memory loads. This paper proposes a low computational cost approach to detect crowd anomalies. The proposed approach avoids the expensive optical flow calculations by adopting a pre-trained 2D convolutional neural network (CNN) for motion information and implements a lighter form of the 2D CNN to achieve high recognition accuracy at low computational cost. Experiments on public datasets show that the proposed model outperforms the existing approaches in terms of detection accuracy alongside providing better performance in generating input frames.

Highlights

Surveillance applications are becoming more important for effective monitoring of crowded places
The results show that the proposed method of using the volumes of interest (VOIs) instead of the entire frame enables the model to learn the features of abnormal behavior better
The set of experiments was concerned with determining the execution times of the stacked grayscale 3-channel image (SG3I) method used in this study with the optical flow [11] and the dynamic image [12]

Summary

Introduction

Surveillance applications are becoming more important for effective monitoring of crowded places. Video surveillance systems must observe abnormal activities involving unusual crowd activity such as crowd chaos, e.g., crowd running in one direction or dispersing from a central point, as well as violent interactions at crowded places, e.g., assault and fighting. There are difficulties in identifying the abnormal behavior itself in scenes containing many people in proximity, as the individuals often appear with high volatility and there is frequent occlusion. Crowd unusual activity is identified based on different parameters, such as its movement pattern and speed, as well as emerging point. The second type of difficulty originates from the high computational complexity involved in most of the abnormal behavior recognition algorithms, which makes it unfeasible to use them in the real world where detection of anomalies in the real-time is required

Methods

Results

Conclusion