摘要
针对现有安全帽佩戴检测中存在的检测精度较低等问题,提出一种基于改进YOLOV5s的安全帽佩戴检测算法。通过结合SimAM注意力,增强安全帽特征的显著性;引入Bi-FPN网络并增加小目标检测层,提高对小目标安全帽的检测精度;采用DIOU-NMS算法提高遮挡目标的检测精度。实验结果证明,改进后的YOLOV5s算法mAP达到97.3%,比原始的YOLOV5s算法提高了4.5%,满足现实场景下安全帽佩戴检测任务的要求。
Aiming at the existing problems such as low detection accuracy in helmet wearing detection,this paper proposes a helmet wearing detection algorithm based on improved YOLOV5s.By combining SimAM attention to enhance the saliency of helmet features;introducing Bi-FPN network and adding small target detection layer the detection accuracy of small target helmets is improved;DIOU-NMS algorithm is used to improve the detection accuracy of obscured targets.The experimental results demonstrate that the improved YOLOV5s algorithm mAP reaches 97.3%,which is 4.5%better than the original YOLOV5s algorithm and meets the requirements of helmet wearing detection tasks in realistic scenarios.
作者
王向前
史策
Wang Xiangqian;Shi Ce(Anhui University of Science and Technology,Huainan Anhui 232001,China)
出处
《山西电子技术》
2023年第6期11-13,38,共4页
Shanxi Electronic Technology