摘要
Abstract
In order to solve the issues of the existing helmet wearing detection algorithms,which have a large number of param-eters,are not conducive to deployment in resource-constrained scenarios,and are prone to false detection and missed detection in complex scenarios,a lightweight helmet wearing detection algorithm of improved YOLOv8 is proposed.On the basis of the original three detection layers of YOLOv8 algorithm,a new small target detection layer is added to enhance the feature extrac-tion ability of the model for small targets.The coordinate attention mechanism is utilized to enhance the YOLOv8 algorithm's focus on key features,thereby improving the algorithm's detection accuracy in complex scenarios.The Ghost-C2f module is de-signed to improve the YOLOv8 algorithm to reduce the number of parameters.The experimental results show that,compared with the YOLOv8n algorithm,the proposed algorithm increases the mean average precision(mAP)by 0.7％and reduces the number of parameters by 39.67％,and has certain advantages compared with other two methods.It can be seen that the pro-posed algorithm not only has higher detection accuracy,but is also more suitable for scenarios with litmited memory resources.关键词
安全帽佩戴检测算法/YOLOv8/坐标注意力机制/Ghost卷积Key words
helmet wearing detection algorithm/YOLOv8/coordinate attention mechanism/Ghost convolution分类
信息技术与安全科学
