基于深度学习的工业生产安全行为分析技术  

Industrial Production Safety Behavior Analysis Technology Based on Deep Learning

作　　者：刘培妍 LIU Peiyan(Yunnan University,Kunming 650206)

机构地区：[1]云南大学,昆明650206

出　　处：《现代制造技术与装备》2022年第11期160-163,共4页Modern Manufacturing Technology and Equipment

摘　　要：随着工业生产建设工作规模的扩大,利用传统的人工监督方式对工业生产行为进行检查,效率低下且易出现监管漏洞。文章提出基于深度学习的工业生产安全行为分析技术,能够自动检测工业生产人员佩戴安全帽的状态、穿戴反光衣的状态以及高中低3种风险级别的区域入侵状态3种工业生产行为。实验结果表明,YOLOv5模型具有较高的检测准确度,基本满足在各种复杂工业生产场景下进行安全行为分析的准确性需求。As the scale of industrial production and construction work increases, the traditional manual supervision to inspect industrial production behaviour is inefficient and prone to regulatory loopholes. Based on this, this paper proposes a deep learning-based industrial production safety behaviour analysis technology, which can automatically detect three industrial production behaviors: the state of industrial production personnel wearing helmets, the state of wearing reflective clothing and the state of intrusion in areas with three risk levels: high, medium and low. The experimental results show that the YOLOv5 model achieves high detection accuracy and meets the accuracy requirements for safety behaviour analysis in various complex industrial production scenarios.

关 键 词：安全帽反光衣检测 危险区域入侵检测 YOLOv5 视频检测 深度学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程] X922.2[环境科学与工程—安全科学]