摘要
火灾是煤矿重大安全事故之一,采用煤矿烟火检测技术是一项有效的预防措施。由于速度、精度和实时性的限制,煤矿烟火检测难以部署到检测系统中,且在小目标上的检测效果较差。因此,本研究提出了一种基于改进的YOLOv5s煤矿烟火检测算法。实验结果表明:在收集的煤矿烟火数据集上,改进的YOLOv5s目标检测算法减少了原网络模型的参数量,有效提升了模型的检测精度与检测速度,小目标的检测效果也得到提升。平均精度提升了2.2%,与原模型相比,每秒的浮点运算数GFLOPs下降40.5%,参数量下降40%,能有效解决煤矿场景下对模型检测的要求。
Fire was one of the major safety accidents in coal mines,and the use of coal mine pyrotechnic detection technology was an effective preventive measure.Due to the limitations of speed,accuracy and real-time,coal mine pyrotechnic detection is difficult to deploy in the detection system and has poor detection effect on small targets.Therefore,this study proposes an improved YOLOv5s coal mine pyrotechnic detection algorithm.Experimental results show that the improved YOLOv5s object detection algorithm reduces the number of parameters of the original network model,effectively improved the detection accuracy and detection speed of the model,and improves the detection effect of small targets on the collected coal mine pyrotechnic dataset.The average accuracy is improved by 2.2 percent,and compared with the original model,the floating point operation GFLOPs per second is reduced by 40.5%,and the number of parameters is reduced by 40%,which can effectively solve the requirements of model detection in coal mining scenarios.
作者
魏少雄
张楠
钟本源
WEI Shaoxiong;ZHANG Nan;ZHONG Benyuan(College of Coal Engineering,Shanxi Datong University,Datong 037003,China;College of Mechanical and Electrical Engineering,Shanxi Datong University,Datong 037003,China)
出处
《煤》
2025年第6期23-26,共4页
Coal
基金
2022年全国煤炭行业教育研究课题(重大课题)(ZMZA20220007)
山西大同大学研究生创新项目(23CX38)。
关键词
YOLOv5s算法
煤矿烟火
精度
参数量
YOLOv5s salgorithm
coal mine pyrotechnics
precision
number of parameters
