测控技术2025，Vol.44Issue(11)：27-36,10.DOI:10.19708/j.ckjs.2025.10.258
基于改进的YOLO-v5对管道高温区域和技术人员检测与识别方法
Detection and Identification Method of Pipeline High-Temperature Zones And Technicians Based on Improved YOLO-v5
摘要
Abstract
The detection of high-temperature zones in pipelines helps to promptly identify and address potential safety hazards.The identification and positioning of technicians can ensure their safety during the work process.Therefore,a method based on improved YOLO-v5 is proposed for precise detection and identification of high-temperature zones and technicians in pipelines.Firstly,a large number of images of high-temperature zones in pipelines and coal mining sites of technicians are collected using infrared cameras,and a dataset is created.Then,a bidirectional feature pyramid network is used to replace the traditional combined feature pyra-mid network(FPN)+path aggregation network(PAN)structure,and an 80 px×80 px high-resolution feature layer is added to enhance the complementarity between shallow details and deep semantic information through cross scale bidirectional feature transfer.A dynamic anchor box density adjustment strategy is introduced in the Head section to alleviate the problem of gradient vanishing for small targets.Experimental results demonstrate that the improved model achieves 95.24%AP@0.5 for pipeline high-temperature zones detection and 90.95%AP@0.5 for technician recognition,which is 5.12%higher than the mean average precision(mAP)of the original YOLO-v5 model,the missed detection rate is reduced by 18.7%in complex scenes such as dust and low light,and it has significant advantages in small target recognition.关键词
管道高温区/改进YOLO-v5/检测/识别Key words
pipeline high-temperature zones/improved YOLO-v5/detection/recognition分类
计算机与自动化引用本文复制引用
李中鹤,李强,赵美蓉..基于改进的YOLO-v5对管道高温区域和技术人员检测与识别方法[J].测控技术,2025,44(11):27-36,10.基金项目
国家重点研发计划(2021YFC2202702) （2021YFC2202702）
