Automatic detection of safety helmet wearing based on head region location

In order to solve the problem of difficult and low precision in detection safety helmet wearing complex pose construction worker, a method based on estimation is proposed. model OpenPose, residual network optimized feature extraction introduced obtain skeletal point information then worker estimated information, three-point localization proposed for front back pose, skin colour side determine head region. The YOLO v4 used detect region, worker's judged according whether region intersects or not. Experimental results show that accuracy higher than other methods, adaptability environment stronger.