Automated On-Site Hazard Identification Using Digital and Thermal Imagery

Abstract: A major component of construction safety management processes entails identifying potentially on-site hazardous areas. However, traditional methods are manual, labor-intensive, and time-consuming. In an attempt to automate hazard identification, researchers have used digital imagery and computer vision at large but failed to explore thermal imaging analysis. As such, this paper introduces an automated vision-based hazard identification approach using both digital and thermal imagery. Several experiments were conducted and results highlighted the effectiveness of analyzing both digital and thermal images to reduce false positive detections and rapidly identify on-site hazardous areas.