基于YOLOv8-OCR的井下人员检测算法  

Underground Personnel Detection Algorithm Based on YOLOv8-OCR

作　　者：倪云峰[1] 霍洁 侯颖[1] 王静[1] 郭苹 NI Yunfeng;HUO Jie;HOU Ying;WANG Jing;GUO Ping(College of Communication and Information Engineering,Xi’an University of Science and Technology,Xi’an 710054,China)

机构地区：[1]西安科技大学通信与信息工程学院,陕西西安710054

出　　处：《无线电工程》2024年第8期1847-1853,共7页Radio Engineering

基　　金：国家自然科学基金(61701393);陕西省教育厅科学研究计划项目(19JK0528,19JK0531)。

摘　　要：为提高井下昏暗环境的目标检测性能,将行人属性应用到井下,将反光号码牌贴在安全帽和工作服上作为属性进行识别。针对井下小目标检测率低的缺点,提出了一种将YOLOv8检测算法与光学字符识别(Optical Character Recognation, OCR)技术相结合的方法。通过添加卷积块注意力模块(Convolutional Block Attention Module, CBAM),提高了网络对不同尺度的特征提取能力;将YOLOv8中的CBS模块改进为CBF模块,提高了目标的识别准确率;对检测到的反光号码牌区域用OCR技术对区域内数字进行识别,进一步提高了模型的检测精度。实验结果表明,提出的方法在自建数据集上获得了93.2%的识别准确率和每张24.4 ms的检测速度,相比YOLOv8模型有着更高的准确率,能够有效地应对井下环境中的光照变化和干扰因素并且满足实时检测的要求。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 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]