欢迎访问《图学学报》 分享到:

图学学报 ›› 2023, Vol. 44 ›› Issue (2): 324-334.DOI: 10.11996/JG.j.2095-302X.2023020324

• 图像处理与计算机视觉 • 上一篇    下一篇

基于改进YOLOv7的X线图像旋转目标检测

成浪1(), 敬超1,2,3()   

  1. 1.桂林理工大学信息科学与工程学院,广西 桂林 541004
    2.桂林理工大学嵌入式技术与智能系统重点实验室,广西 桂林 541004
    3.桂林电子科技大学可信软件重点实验室,广西 桂林 541004
  • 收稿日期:2022-09-28 接受日期:2022-11-08 出版日期:2023-04-30 发布日期:2023-05-01
  • 通讯作者: 敬超(1983-),男,副教授,博士。主要研究方向为机器学习、图像处理。E-mail:jingchao@glut.edu.cn
  • 作者简介:成浪(1995-),男,硕士研究生。主要研究方向为计算机视觉、图像处理。E-mail:862409782@qq.com
  • 基金资助:
    国家自然科学基金项目(61802085);国家自然科学基金项目(61862019);广西自然科学基金项目(2020GXNSFAA159038);广西可信软件重点实验室基金项目(kx202011);广西中青年教师基础能力提升项目(2022KY0252)

X-ray image rotating object detection based on improved YOLOv7

CHENG Lang1(), JING Chao1,2,3()   

  1. 1. School of Information Science and Engineering, Guilin University of Technology, Guilin Guangxi 541004, China
    2. Guangxi Key Laboratory of Embedded Technology and Intelligent System, Guilin University of Technology, Guilin Guangxi 541004, China
    3. Guangxi Key Laboratory of Trusted Software, Guilin University of Electronic Science and Technology, Guilin Guangxi 541004, China
  • Received:2022-09-28 Accepted:2022-11-08 Online:2023-04-30 Published:2023-05-01
  • Contact: JING Chao (1983-), associate professor, Ph.D. His main research interests cover machine learning and image processing. E-mail:jingchao@glut.edu.cn
  • About author:CHENG Lang (1995-), master student. His main research interests cover computer vision and image processing. E-mail:862409782@qq.com
  • Supported by:
    National Natural Science Foundation of China(61802085);National Natural Science Foundation of China(61862019);Guangxi Natural Science Foundation(2020GXNSFAA159038);Guangxi Trusted Software Key Laboratory Fund(kx202011);Guangxi Middle-Aged and Young Teachers′ Basic Ability Improvement Project(2022KY0252)

摘要:

针对X线图像违禁品目标检测中存在的识别定位困难以及忽略物品方向性的问题,提出了一种基于改进YOLOv7的X线图像旋转目标检测算法。首先,通过在原网络中融合高效注意力机制模块提高模型对深层重要特征的提取能力;然后,改进扩展的高效长程注意力机制的特征融合路径,在模块之间增加跳跃连接和1×1卷积架构,使网络提取更丰富的物品特征;最后,针对X线图像中违禁品放置方向任意的问题,使用密集编码标签表示法对角度进行离散化处理,提高违禁品定位的准确性。实验结果表明,改进的算法在HiXray,OPIXray和PIDray数据集上分别取得了91.2%,92.6%和66.4%的检测精度,较原YOLOv7模型分别提高了20.2%,10.6%和15.5%,在有效提高X线图像违禁品检测精度的基础上,为保障公共安全提供了很好的技术支持。

关键词: 旋转目标检测, 注意力机制, X线图像, YOLOv7, 违禁品

Abstract:

For prohibited items in X-ray images, an algorithm for the detection of rotating targets based on the improved YOLOv7 was proposed to address the challenges of accurate identification and localization, as well as the neglection of the directionality of the items. Firstly, an efficient attention network module was integrated into the original network to enhance the ability of the model to extract deep important features. Then, the feature fusion path of the extended efficient long-range attention network (E-ELAN) was improved, and the residual structure jump connection and 1×1 convolution were added between modules, allowing the network to extract richer item features. Finally, to tackle the problem of arbitrary placement direction of prohibited items in X-ray images, the angles were discretized using the dense coded label representation method, thereby improving the positioning accuracy of prohibited items. The experimental results revealed that the improved algorithm could achieve a detection accuracy of 91.2%, 92.6%, and 66.4% on HiXray, OPIXray, and PIDray datasets, respectively. Compared with the original YOLOv7 model, the results were improved by 20.2%, 10.6%, and 15.5%, respectively. The proposed algorithm could provide a valuable technical support for public security by effectively improving the accuracy of prohibited item detection in X-ray images.

Key words: rotating target detection, attention mechanism, X-ray images, YOLOv7, prohibited item

中图分类号: