Wearing Safety Helmet Detection in Substation
Zhong Kai, Wang Xiaozhi · 2019
Wearing safety helmet is very essential for the safety of staff of power substations. In this paper, it proposes a detection framework of safety helmet based on the theory of machine learning and image processing technology. TensorFlow model enables to detect pedestrians, while the head position will be detected by the analysis of human Head-to-Body Ratio. Then the color space transformation and the color feature discrimination of head position are used to detect whether the staff wearing safety helmets or not. The experimental results of a large number of substation monitoring video sequences have shown that the recognition rate of wearing safety helmet detection system reach 89.0%, which proves the effectiveness of the proposed framework.