无线电工程2024，Vol.54Issue(8)：1847-1853,7.DOI:10.3969/j.issn.1003-3106.2024.08.002
于YOLOv8-OCR的井下人员检测算法
Underground Personnel Detection Algorithm Based on YOLOv8-OCR
摘要
Abstract
In order to improve the target detection performance in dark underground environment,the pedestrian attributes is innovatively applied to the underground environment,and the reflective number plate is pasted on the helmet and work clothes as the attributes for recognition.To address the disadvantage of low detection rate of small targets,a method combining YOLOv8 detection algorithm and Optic al Character Recognition(OCR)technology is proposed.By adding a Convolutional Block Attention Module(CBAM),the feature extraction capability of different scales is improved;the CBS module in YOLOv8 is modified to CBF module to improve the target recognition accuracy;then the number in the detected reflective number plate area is identified by OCR technology,which further improves the detection accuracy of the model.The experimental results show that the proposed method achieves an recognition accuracy of 93.2％,which is higher than that of YOLOv8 model,and a detection speed of 24.4 ms per piece on the self-built data set.The method can effectively deal with the light changes and interference factors in downhole environment and meet the requirements of real-time detection.关键词
目标检测/YOLOv8/光学字符识别/反光号码牌/注意力机制Key words
target detection/YOLOv8/OCR/reflective number plate/attention mechanism分类
信息技术与安全科学引用本文复制引用
倪云峰,霍洁,侯颖,王静,郭苹..于YOLOv8-OCR的井下人员检测算法[J].无线电工程,2024,54(8):1847-1853,7.基金项目
国家自然科学基金(61701393) （61701393）
陕西省教育厅科学研究计划项目(19JK0528,19JK0531)National Natural Science Foundation of China(61701393) （19JK0528,19JK0531）
Scientific Research Program of Shaanxi Provincial Depart-ment of Education(19JK0528,19JK0531) （19JK0528,19JK0531）
