【摘要】 建筑行业是一种高危行业。在建筑安全中，安全帽的佩戴可以在一定程度上保障施工人员的安全。针对施工人员的安全帽佩戴问题，设计并实现了一款基于YOLOv5s算法模型的智慧工地安全管理系统。将训练好的算法模型通过RT-Thread操作系统部署于嵌入式硬件平台，在施工现场智能识别未佩戴安全帽的人员并提出报警。在进行推理测试后得出结果，基于YOLOv5s的算法模型可以有效地区别出施工人员有无正确佩戴安全帽，测试精度达到92.3%。当IoU为50时，mAP值达到93.1%。实验结果表明，基于YOLOv5s的算法模型在人群密集和小头检测等问题上准确率高，实时性强，均已达到实际使用需求，同时有助于降低施工风险，减少不必要的人力监督，实现工地人员智能安全管理。更多还原

【Abstract】 The construction industry is a high-risk industry, and the wearing of safety helmets by construction workers can ensure their safety to some extent. In response to the issue of safety helmet wearing by construction workers, a smart construction site safety management system based on the YOLOv5s algorithm model was designed and implemented. The trained algorithm model was deployed on an embedded hardware platform through the RT-Thread operating system, which can intelligently recognize personnel who are not wearing safety helmets at the construction site and issue an alarm. After inference testing, it was found that the YOLOv5s-based algorithm model can effectively distinguish whether construction workers are wearing safety helmets correctly, with a test accuracy of 92.3%. Under the condition of IoU=50, the mAP value reaches 93.1%. Based on the experimental results, the YOLOv5s-based algorithm model has high accuracy in problems such as crowded people and small head detection, strong real-time performance, and has met practical usage requirements. It can also help reduce construction risks, reduce unnecessary manual supervision, and achieve intelligent safety management of construction site personnel.更多还原