Safety Helmet Wearing Detection and Recognition Based on YOLOv4
Mudi Zhou, Zhuli Fang, Bin Zhao, Pengfei Li · 2021 3rd International Academic Exchange Conference on Science and Technology Innovation (IAECST) · 2021
In specific working conditions, construction personnel need to wear safety helmets. Safety hazards occur when workers or other people enter the site without helmets. In order to monitor whether operators meet safety standards, we improved YOLOv4 algorithm to detect and identify whether people entering special places are wearing safety hats. We improve the loss function of the convolutional neural network and change its attention mechanism to make the network more sensitive to smaller objects. Moreover, we added some parameters to solve the imbalance between classes. Experimental results show that our improvement improves the accuracy of the network model in detecting and identifying the wearing condition of safety helmet.