基于改进RT-DETR的车门内拉手表面缺陷检测方法  

Surface defect detection method for inner handle of car door based on improved RT-DETR

作　　者：徐仟祥 曾勇[1] 卢倩[1] 南玉龙 Xu Qianxiang;Zeng Yong;Lu Qian;Nan Yulong(College of Mechanical Engineering,Yancheng Institute of Technology,Yancheng 224000,China)

机构地区：[1]盐城工学院机械工程学院,盐城224000

出　　处：《电子测量技术》2024年第18期172-181,共10页Electronic Measurement Technology

基　　金：江苏省高等学校基础科学(自然科学)基金(21KJA460009);江苏省高校自然科学基金面上项目(22KJD460009);江苏省高等学校基础科学(自然科学)研究面上项目(23KJD460009)资助。

摘　　要：针对车门内拉手表面的缺陷目标小、多尺度、易反光等问题。首先,通过使用碗状光源和降低图像采集表面夹角的方法,解决内拉手表面图像采集时因表面弯曲和镜面反射导致的缺陷特征被覆盖问题。然后,针对传统的RT-DETR模型存在缺陷检测精度差,速度慢等问题,提出一种改进的RT-DETR目标检测方法。该方法首先以RT-DETR为基础架构,在主干网络中采用并行的膨胀卷积与CA注意力机制并结合卷积重参数化的方式,以增加网络感受野和建立长距离的语义信息的同时提高网络推理速度。其次,通过添加额外的检测层来增加网络对小目标检测的特征提取能力。紧接着,在多尺度特征融合阶段使用了改进的BIFPN结构以提高模型信息交互的能力。最后,消融实验表明,相较于传统的基于RT-DETR的检测方法,本文提出的改进RT-DETR的检测方法,平均精度提升了6.5%,检测速度为传统模型的1.6倍,同时模型的参数量仅为原网络的76.5%。验证了本文所提方法的有效性。To address the challenges of small defect targets,multi-scale issues,and high reflectivity on the surface of the inner car door handle,we first tackle the problem of defect features being obscured during image acquisition due to surface curvature and mirror reflection by using a bowl-shaped light source and reducing the angle of the image acquisition surface.Then,recognizing the limitations of traditional RT-DETR models,such as poor detection accuracy and slow speed,we propose an improved RT-DETR object detection method.This method builds upon the RT-DETR framework,utilizing parallel dilated convolutions and the CA attention mechanism combined with convolutional re-parameterization in the backbone network to increase the receptive field and establish long-distance semantic information while improving the network inference speed.Additionally,extra detection layers are added to improve the network′s feature extraction capability for small object detection.In the multi-scale feature fusion stage,we use an improved BIFPN structure to enhance the model′s information interaction capability.Finally,ablation experiments show that,compared to traditional RT-DETR-based detection methods,our proposed improved RT-DETR method increases the mean Average Precision by 6.5%,achieves a detection speed 1.6 times that of the traditional model,and reduces the model′s parameter count to only 76.5%of the original network,validating the effectiveness of our proposed method.

关 键 词：车门内拉手 缺陷检测 RT-DETR 膨胀卷积 卷积重参数化 BIFPN 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP29[自动化与计算机技术—计算机科学与技术] TN0[电子电信—物理电子学]