Computer Vision based System for Fall Detecting of Construction Workers
Abstract
While construction projects play a significant role in growing the economy of the societies by creating jobs, developing infrastructures, etc., accidents can be the downside of these projects. In addition to the direct... [ view full abstract ]
While construction projects play a significant role in growing the economy of the societies by creating jobs, developing infrastructures, etc., accidents can be the downside of these projects. In addition to the direct physical and financial impacts of the accidents, they also have negative impact on the morale of project participants. Safety regulations aim to prevent or reduce construction accidents. However, statistics still show a large number of accidents. Therefore, investing in proactive safety solutions is important. Furthermore, detecting an accident while it is happening can also mitigate the impact of that accident by providing immediate assistance to the victims. This paper proposes a fall detection system using computer vision to catch the fall accident of construction workers in near real-time, so that the worker can be found and rescued as soon as possible. The proposed method includes an offline training module and an online detection module. In the offline training module, multiple classifiers are trained and compared using the features of the worker’s binary segment. Then, in the online detection module, the trained classifiers are applied on the new binary images of the workers in order to detect a potential fall. Moreover, sensitivity analysis is performed to study the effect of the number of sequential frames used for classifying the worker body movements on the accuracy of the classifiers.
Authors
-
Mohammad Soltani
(Concordia University)
-
Walid Aly
(indus.ai)
-
Zhenhua Zhu
(Concordia University)
-
Amin Hammad
(Concordia University)
Topic Areas
Automation and robotics for construction , Big Data, data mining and machine learning
Session
O10 » Health and Safety (12:45 - Wednesday, 6th June, Sonaatti 2)
Paper
full_paper_icccbe2018_v4.pdf
Presentation Files
The presenter has not uploaded any presentation files.