A Dual Transformer-Based Deep Learning Model for Passenger Anomaly Behavior Detection in Elevator Cabs

A Dual Transformer-Based Deep Learning Model for Passenger Anomaly Behavior Detection in Elevator Cabs

Yijin Ji (Special Equipment Safety Supervision Inspection Institute of Jiangsu Province, China), Haoxiang Sun (Special Equipment Safety Supervision Inspection Institute of Jiangsu Province, China), Benlian Xu (Suzhou University of Science and Technology, China), Mingli Lu (Changshu Institute of Technology, China), Xu Zhou (Changshu Institute of Technology, China), and Jian Shi (Changshu Institute of Technology, China)
Copyright: © 2024 |Pages: 14
DOI: 10.4018/IJSIR.361578
Article PDF Download
Open access articles are freely available for download

Abstract

Effective detection of abnormal behaviors within elevator cabins is critical to ensure elevator safety. While existing deep learning based anomaly detection methods mainly focus on convolutional neural networks for spatial feature extraction and recurrent networks for temporal feature learning, recent advancements in the Transformer architecture have demonstrated its power in time series predictions, and extended its capabilities to vision detection tasks. In this study, we present a duel transformer-based framework that can proficiently detect falling and fighting events in elevator cabs. The proposed solution leverages the vision transformer (ViT) to extract frame-level spatial features, followed by a temporal Transformer to identify abnormalities in surveillance videos. A comprehensive comparison between the proposed transformer-based method and other traditional recurrent neural network variants is carried out to validate the effectiveness of the method.
Article Preview
Top

Introduction

China has become the world’s leading country in elevator production and sales, a testament to its rapid urbanization and high-rise building development. According to public data released by the State Administration for Market Regulation, the number of elevators in China reached a staggering 9.6446 million by the end of 2022. Elevators have become a major vertical transportation facility, significantly enhancing convenience in people’s daily lives. However, despite their widespread use, elevator cabs have raised safety concerns due to their enclosed nature, isolated from the outside world. Two primary issues have come to the forefront. Firstly, they present a risk of accidents, particularly for vulnerable individuals such as the elderly, the weak, the sick, or children, who may be at higher risk of falling when traveling alone in the elevator cab. Secondly, the confined space within elevators may offer opportunities for criminal activities, including robbery, kidnapping, violence, and other illicit acts.

In order to ensure a safe environment within elevator cabs, it is common practice to install cameras in the top corners to monitor passenger behavior. Currently, in most public places such as hotels, office buildings, hospitals, schools, and train stations, elevator monitoring still relies on manual surveillance. This traditional method has several issues and drawbacks. Firstly, if surveillance operators are tasked with monitoring multiple screens simultaneously, it can lead to fatigue and reduced attention spans. This makes it challenging to achieve 24/7 real-time monitoring, resulting in a potential loss of valuable information. Secondly, traditional monitoring systems merely record and save surveillance videos passively. In the event of an incident, operators must manually search through a massive amount of recorded footage to obtain evidence, which is a time-consuming process. Finally, abnormal behaviors in elevator cabs are relatively rare compared to the normal behaviors exhibited by passengers. If the surveillance system fails to respond to abnormal behaviors promptly and issue an alarm, it can lead to irreparable losses and pose a serious threat to social harmony and stability. Despite advancements in technology, video surveillance systems still rely on human operators due to the presence of false positives, which makes the task complex and challenging. As a result, there is a growing demand for intelligent and robust video surveillance systems capable of automatically detecting abnormal behaviors and raising alarms. In recent years, researchers have conducted extensive studies on various abnormal passenger behaviors in elevator cabs, including falling, jumping, fighting, and door jamming (Lan & Li, 2021; Liu et al., 2021; Roka et al., 2023; Shi et al., 2021; Shu et al., 2014a; Zhu & Wang., 2016). In this work, we aim to push the boundaries beyond conventional anomaly detection methods that merely distinguish abnormal behaviors from normal events.

Complete Article List

Search this Journal:
Reset
Volume 16: 1 Issue (2025)
Volume 15: 1 Issue (2024)
Volume 14: 3 Issues (2023)
Volume 13: 4 Issues (2022)
Volume 12: 4 Issues (2021)
Volume 11: 4 Issues (2020)
Volume 10: 4 Issues (2019)
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing