Skeleton-based automatic assessment and prediction of intrusion risk in construction hazardous areas

Intrusion behavior in hazardous areas is one of the major causes of construction safety accidents including falls from height, strikes by objects, etc. Implementing automatic and precise assessment of intrusions to enhance safety performance is of great importance in construction areas. Due to the large area of construction sites and diverse human behaviors, it is difficult to accurately predict worker behavior, resulting in many intrusions detected after the occurrence. Notably, computer vision-based skeleton extraction can provide a promising noncontact solution for assessing intrusions. This paper presents a novel intrusion behavior detection and evaluation approach by defining a safety buffer zone and using two key quantitative elements, i.e. the motion state and orientation posture of intruders. An indoor experiment was conducted by employing skeleton detection technology with safety knowledge to demonstrate the feasibility and effectiveness of the assessment method. The participants' risk levels were evaluated separately and simultaneously based on the motion and posture. The risk level was compared based on various evaluated methods and the ground truth. The results show that a satisfying accuracy of intrusion assessment can be achieved at different risk levels. Appropriate warning and intervention methods can be implemented to mitigate the occurrence or reduce the severity of intrusions and thus reduce safety accidents with the use of the proposed method.