Fall Prevention from Scaffolding Using Computer Vision and IoT-Based Monitoring
- Authors
- Khan, Muhammad; Khalid, Rabia; Anjum, Sharjeel; Tran, Si Van-Tien; Park, Chansik
- Issue Date
- Jul-2022
- Publisher
- American Society of Civil Engineers (ASCE)
- Keywords
- Accidents; Computer vision; Construction worker safety; Fall from height (FFH); Internet-of-Things (IoT)-based monitoring; Scaffold; Sensors; Smart safety hook (SSH)
- Citation
- Journal of Construction Engineering and Management, v.148, no.7
- Journal Title
- Journal of Construction Engineering and Management
- Volume
- 148
- Number
- 7
- URI
- https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/57801
- DOI
- 10.1061/(ASCE)CO.1943-7862.0002278
- ISSN
- 0733-9364
1943-7862
- Abstract
- Fall from height (FFH) is the most significant cause of occupational fatalities in the construction industry, accounting for approximately 54% of all accidents. Such fatalities have decreased considerably due to the use of personal protective equipment (PPE). However, the manual monitoring of compliance to PPE is complex and challenging for site managers. Automation in construction safety presents multiple solutions for monitoring safety at sites. In this study, a smart safety hook (SSH) monitoring method is proposed to eliminate the risk associated with FFH accidents by integrating computer vision [closed-circuit TV (CCTV)-imagery] and Internet-of-Things (IoT)-based [inertial measurement unit (IMU)IMU and altimeter] monitoring technologies. The proposed monitoring approach is validated through five real-time scenarios: (1) attached to the scaffolding and h>1.82 m (6 ft), (2) attached to the worker and h>1.82 m, (3) unattached and h>1.82 m, (4) h<1.82 m, and (5) outside of the risk zone. The proposed technique aims to relieve the site manager's or safety engineer's workload by smartly and instantaneously alerting of workers' unsafe behavior (via alarm, LED blinking, and bounding box on live camera feed). Moreover, the IoT-based hardware setup goes to low power to extend the battery life when there is no unsafe behavior. The experimental results demonstrate that the proposed solution exhibits more than 98% accuracy for real-time detection and classification. Furthermore, it can be extended to monitor several workers and their location data in the future. © 2022 American Society of Civil Engineers.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - ETC > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/57801)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.