A Detection Method of Unsafe Behavior in Substation Based on Deep Learning
Mingzeng Zhu, Qiuqin Qin, Chengwei Huang, Wei Zhang, Zhaoting Liang, Jiwan Chen · Proceedings of the 3rd International Conference on Information Technologies and Electrical Engineering · 2020
During the operation of the substation, the irregular and unsafe operation of the staff will bring safety hazards to the stable operation of the power grid, threaten the safety of the staff, and may cause catastrophic consequences. Therefore, it is very important to detect and identify the behavior of substation workers. Based on this, this paper mainly studies the detection and recognition methods of two unsafe behaviors of substation workers not wearing safety helmets and smoking in the workplace. The deep learning method is used in this paper to improve the deep learning capabilities of the learning deeply supervised object detectors (DSOD) from scratch detector, so that it can detect the two unsafe behaviors mentioned above while considering the same detection object. Traditional recognition methods are easy to misjudge pictures and videos, and the detection method based on deep learning proposed in this paper has the advantage of accurate and reliable recognition effect compared with traditional recognition methods.