Open AccessResearch on Wear Recognition of Electric Worker’s Helmet Based on Neural Network
Author(s) -
Yuwei Sun,
Jing Fu,
Qingshan Ma,
Xinghua Yu,
Guoxiang Fan
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1449/1/012057
Subject(s) - computer science , artificial neural network , kalman filter , scheme (mathematics) , artificial intelligence , power (physics) , identification (biology) , video monitoring , electric power , computer vision , real time computing , mathematical analysis , physics , botany , mathematics , quantum mechanics , biology
Head injury is an important cause of electric construction accidents. Wearing a safety helmet is an effective measure to prevent electric construction accidents, and in the work, unsafe behavior of workers without wearing a helmet often occurs. This paper proposes a neural network-based automatic cap wear automatic identification and detection technology to identify the behavior of unwearing helmets in power construction. In view of the shortcomings of the existing helmet detection scheme in the complex environment, such as low efficiency and low recognition accuracy, this paper selects the region of the neural network (Fast-er-RCNN) to identify the wearing of the helmet during the power operation. The overall scheme uses the target detection scheme based on the foreground detection technology to detect the target individual in the obtained video recording of the power operation. The Kalman filter algorithm is used to track the target in the video and solve the occlusion problem in the video image. Through the actual detection of power construction monitoring video, the availability and efficiency of the designed algorithm are verified. The target recognition accuracy can reach 85.7% and the recall rate is 87.5%.