为解决危大工程中吊装作业安全管理的问题,基于深度学习构建目标检测算法(You Only Look Once version 5,YOLOv5)网络模型,针对进入吊装作业区域内人员的防护装备进行多目标融合检测,并对吊钩在施工过程中的状态进行检测。在原始的检测...为解决危大工程中吊装作业安全管理的问题,基于深度学习构建目标检测算法(You Only Look Once version 5,YOLOv5)网络模型,针对进入吊装作业区域内人员的防护装备进行多目标融合检测,并对吊钩在施工过程中的状态进行检测。在原始的检测网络模型中引入4种注意力机制,并通过5种训练模型的结果对比分析,进而选择卷积块注意力模块(Convolutional Block Attention Module,CBAM)最优模型。优化后的检测模型对安全帽的平均识别精度达86.5%,对反光衣的平均识别精度达83.0%,对吊钩的状态识别精度达92.0%。将训练好的人员检测模型和吊钩检测模型打包成exe执行文件,应用到施工安全管理人员的中控平台,可帮助管理人员更好地判断吊装作业的工作情况,进而及时进行风险管控。展开更多
Security experts have not formally defined the distinction between viruses and normal programs. The paper takes user＇s intension as the criteria for malice, gives a formal definition of viruses that aim at stealing o...Security experts have not formally defined the distinction between viruses and normal programs. The paper takes user＇s intension as the criteria for malice, gives a formal definition of viruses that aim at stealing or destroying files, and proposes an algorithm to detect virus correctly. Compared with traditional definitions, this new definition is easy to understand, covers more malwares, adapts development of virus technology, and defines virus on the spot. The paper has also analyzed more than 250 real viruses and finds that they are all in the domain of the new definition, this implies that the new definition has great practical significance.展开更多
文摘为解决危大工程中吊装作业安全管理的问题,基于深度学习构建目标检测算法(You Only Look Once version 5,YOLOv5)网络模型,针对进入吊装作业区域内人员的防护装备进行多目标融合检测,并对吊钩在施工过程中的状态进行检测。在原始的检测网络模型中引入4种注意力机制,并通过5种训练模型的结果对比分析,进而选择卷积块注意力模块(Convolutional Block Attention Module,CBAM)最优模型。优化后的检测模型对安全帽的平均识别精度达86.5%,对反光衣的平均识别精度达83.0%,对吊钩的状态识别精度达92.0%。将训练好的人员检测模型和吊钩检测模型打包成exe执行文件,应用到施工安全管理人员的中控平台,可帮助管理人员更好地判断吊装作业的工作情况,进而及时进行风险管控。
基金Supported by the Foundation of National Labora-tory for Modern Communications(51436050505KG0101)
文摘Security experts have not formally defined the distinction between viruses and normal programs. The paper takes user＇s intension as the criteria for malice, gives a formal definition of viruses that aim at stealing or destroying files, and proposes an algorithm to detect virus correctly. Compared with traditional definitions, this new definition is easy to understand, covers more malwares, adapts development of virus technology, and defines virus on the spot. The paper has also analyzed more than 250 real viruses and finds that they are all in the domain of the new definition, this implies that the new definition has great practical significance.
