Development of a safety monitoring system in building construction

Abstract

Long term impact from sustaining back injury is expected. Not only it could affect the performance of work in the future, it could also get severe as one gets older. Getting back injury in the construction industry is quite common, considering that back injury has been the top contributor to the musculoskeletal disorder cases in Singapore for the past 3.5 years. With the aim of reducing such cases, one of deep learning’s architecture, Convolutional Neural Network (CNN), is then introduced. CNN has achieved great results in object detection and image recognition. Therefore, a form of CNN, OpenPose was developed to perform the angle detection of human posture, particularly in construction industry during manual lifting. Another form of CNN, You Only Look Once (YOLO) was also used to detect the usage of mask. In which, at the end of the project, these 2 CNN’s forms, OpenPose and YOLO will be integrated to perform mask as well as angle detection.