건설현장에서의 컴퓨터 비전 기반 안전모 자동인식에 관한 기초연구

Abstract

A safety helmet is a personal protective device (PPE) that is designed to protect the workers’ head from accidents at industrial sites, particularly at the highly risky places such as the construction sites. It is regulated to wear a safety helmet. However, despite of the regulation, there are many cases not wearing helmets because of their discomforts, which results in critical disasters. It is not only necessary to prevent workers from being injured by wearing safety helmets at construction sites, but also to activate practical practices to prevent the accidents from being caused by the failure to wear them during the year. On the other hand, there are many investigations with the core elements of the 4th Industrial Revolution, which enables automatic recognition objects or people with computerized vision technology and video imaging devices such as cameras. In particular, a recent safety study was actively conducted in the construction fields to automatically recognize and trace workers without wearing safety helmets at project sites by CCTV cameras or cameras installed at construction sites. In this study, we will explore the applicability of the machine-running techniques with the development of computerized vision - based automatic recognition safety helmets in various forms and conditions to detect the best recognition rate of safety helmets among of the combination of multiple feature vectors with the images on efficiency of recognition taken from a real construction site.