A BIM Integrated Hazardous Zone Registration Using Image Stitching

Identifying hazardous zones is one of the priorities and duties of construction safety monitoring. With the emergence of vision intelligence technology, hazardous zones detected by AI algorithms can support safety managers in predicting potential hazards as well as making decisions. However, the risky area in images/videos continues to suffer from numerous issues as undetected or low accuracy. Further, there are many kinds of hazardous zones that need to be trained for computer vision. This study aims to develop a BIM integrated application of hazardous zone registration for the construction stage, which is represented as construction objects using image stitching. The methodology consists of two stages: (1) during the pre-construction stage, BIM-based safety planning extracts the virtual image with the danger region; and (2) during the construction stage, the hazardous zone may be recorded in the site image using a stitching algorithm. A proposed method has been validated by visualizing hazardous zones from the actual images of the ringlock scaffold systems installed at the ConTIL laboratory. As a result, the BIM integrated application is anticipated to assist vision intelligent warning systems in proactively predefining hazardous zones, ultimately reducing repeated accidents associated with other entities entering dangerous areas.