基于深度学习的消防器材自动识别研究

Research on automatic identification of fire fighting equipment based on deep learning

在智慧消防城市救援平台项目中,定位建筑内消防器材位置的前提是准确识别建筑图纸上消防器材图标,只有精准地识别建筑图纸上的消防器材,救援平台才能为消防救援行动提供有效支撑。针对手动录入建筑楼层消防信息效率低下的问题,文中提出了一种在darknet框架下利用YOLOv3算法来实现的基于深度学习的消防器材自动识别方案。通过收集数据集,下载预训练文件,使用YOLOv3算法进行自训练的方法,达到在消防器材图标数量和种类众多的建筑图纸上实现对消防器材图标准确识别与位置输出的目的。实验结果表明,消防器材自动识别方案能显著提高智慧消防城市救援平台项目中建筑图纸上消防器材图标录入的效率,具有很强的可靠性。

In the smart fire city rescue platform project,the premise of locating the location of fire⁃fighting equipment in the building is to identify the fire⁃fighting equipment on the architectural drawings.Only by accurately identifying the fire equipment on the architectural drawings can the rescue platform provide effective support for fire rescue operations.Aiming at the inefficiency of manually inputting fire information on building floors,a deep learn⁃based automatic fire equipment identification scheme is proposed based on YOLOv3 algorithm under darknet framework.By collecting data sets,downloading pre⁃training files,and using YOLOv3 algorithm to conduct self⁃training,the purpose of accurate identification and location output of fire equipment icons can be achieved on the numerous and varied architectural drawings of fire equipment icons.The experimental results show that the scheme of automatic identification of fire equipment can significantly improve the efficiency of inputting fire equipment icons on architectural drawings of intelligent fire fighting urban rescue platform project and has strong reliability.