RGE-YOLO：下采样与注意力机制融合的缺陷检测方法

doi:10.3778/j.issn.1002-8331.2505-0255

摘要/Abstract

摘要： 针对工业缺陷检测任务中精度与效率难以兼顾的问题，构建了一种基于YOLOv8n的改进缺陷检测算法RGE-YOLO。为减少传统卷积下采样导致的小目标特征丢失，该算法引入鲁棒特征下采样（robust feature downsampling，RFD）模块，以多尺度特征融合机制增强缺陷的表征能力。同时，为提升模型对缺陷区域的关注度，设计了轻量化注意力模块（ghost-simple attention module，GhostSAM），其融合了Ghost卷积的高效特征提取能力与无参数注意力的空间激活特性，并结合门控策略实现通道语义的自适应调节。此外，采用高效交并比（efficient intersection over union，EIoU）损失函数优化边界框回归过程。在NEU-DET钢材缺陷数据集上的实验结果表明，与基线算法YOLOv8n相比，RGE-YOLO算法的精确度、召回率和mAP@0.5分别提升了8.3、2.7和3.8个百分点，另外改进后网络的检测速度为131帧/s。实验结果表明，该算法实现了检测精度与效率的有效平衡。

关键词: 缺陷检测, YOLOv8, 注意力机制, 多尺度特征融合

Abstract: To address the trade-off between accuracy and efficiency in industrial defect detection, an improved algorithm, RGE-YOLO, is proposed based on YOLOv8n. The algorithm introduces a robust feature downsampling (RFD) module to mitigate the small-target feature loss caused by conventional convolutional downsampling, enhancing feature representation through a multi-scale fusion mechanism. Concurrently, a lightweight attention module, the ghost-simple attention module (GhostSAM), is designed to improve the model??s focus on defect regions. It integrates the efficient feature extraction of Ghost convolution with the spatial activation of a parameter-free attention mechanism, and incorporates a gating strategy for adaptive channel-wise semantic adjustment. Furthermore, the bounding box regression process is optimized using the efficient intersection over union (EIoU) loss function. Experimental results on the NEU-DET steel defect dataset show that, compared to the baseline YOLOv8n, RGE-YOLO increases precision, recall, and mAP@0.5 by 8.3, 2.7, and 3.8 percentage points, respectively, while achieving a detection speed of 131 frames per second. These results confirm that the algorithm effectively balances detection accuracy and efficiency.

Key words: defect detection, YOLOv8, attention mechanism, multi-scale feature fusion

汪兴, 韩志科, 黄晓辉, 刘鹏. RGE-YOLO：下采样与注意力机制融合的缺陷检测方法[J]. 计算机工程与应用, 2026, 62(8): 105-115.

WANG Xing, HAN Zhike, HUANG Xiaohui, LIU Peng. RGE-YOLO: Downsampling-Attention Fusion Approach for Defect Detection[J]. Computer Engineering and Applications, 2026, 62(8): 105-115.

参考文献

[1] SABERIRONAGHI A，REN J，EI-GINDY M.Defect detection methods for industrial products using deep learning techniques：a review[J].Algorithms，2023，16（2）：95.
[2] REN Z H，FANG F Z，YAN N，et al.State of the art in defect detection based on machine vision[J].International Journal of Precision Engineering and Manufacturing-Green Technology，2022，9（2）：661-691.
[3] CHEN C，ZHANG X Y.Research on laser ultrasonicsurface defect identification based on a support vector machine[J].Science Progress，2021，104（4）：1-10.
[4] MA J J，HU S Y，FU J Z，et al.A hierarchical attention detector for bearing surface defect detection[J].Expert Systems with Applications，2024，239：122365.
[5] LI X S，ZHU Y.A real-time and accurate convolutional neural network for fabric defect detection[J].Complex & Intelligent Systems，2024，10（3）：3371-3387.
[6] SU P，HAN H Z，LIU M，et al.MOD-YOLO：rethinking the YOLO architecture at the level of feature information and applying it to crack detection[J].Expert Systems with Applications，2024，237：121346.
[7] 高磊，徐洪，汪昌平.基于改进Yolov3的H型钢喷号识别系统研究[J].冶金动力，2024（3）：63-66.
GAO L，XU H，WANG C P.Research on H-beam spray number identification system based on improved YOLOv3[J].Metallurgical power，2024，（3）：63-66.
[8] WANG L，LIU X B，MA J T，et al.Real-time steel surface defect detection with improved multi-scale YOLO-v5[J].Processes，2023，11（5）：1357.
[9] 卢俊哲，张铖怡，刘世鹏，等.面向复杂环境中带钢表面缺陷检测的轻量级DCN-YOLO[J].计算机工程与应用，2023，59（15）：318-328.
LU J Z，ZHANG G Y，LIU S P，et al.Lightweight DCN-YOLO for strip surface defect detection in complex environments[J].Computer Engineering and Applications，2023，59（15）：318-328.
[10] AMERI R，HSU C C，BAND S S.A systematic review of deep learning approaches for surface defect detection in industrial applications[J].Engineering Applications of Artificial Intelligence，2024，130：107717.
[11] GIRSHICK R.Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision.Piscataway：IEEE，2015：1440-1448.
[12] REDMON J，DIVVALA S，GIRSHICK R，et al.You only look once：unified，real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Piscataway：IEEE，2016：779-788.
[13] LIU W，ANGUELOV D，ERHAN D，et al.SSD：single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision.Cham：Springer，2016：21-37.
[14] 伍麟，郝鸿宇，宋友.基于计算机视觉的工业金属表面缺陷检测综述[J].自动化学报，2024，50（7）：1261-1283.
WU L，HAO H Y，SONG Y.A review of metal surface defect detection based on computer vision[J].Acta Automatica Sinica，2024，50（7）：1261-1283.
[15] HU J，SHEN L，SUN G.Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Piscataway：IEEE，2018：7132-7141.
[16] WANG Q L，WU B G，ZHU P F，et al.ECA-Net：efficient channel attention for deep convolutional neural networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway：IEEE，2020：11534-11542.
[17] WOO S，PARK J，LEE J Y，et al.CBAM：convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision（ECCV）.Cham：Springer，2018：3-19.
[18] HOU Q B，ZHOU D Q，FENG J S.Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway：IEEE，2021：13713-13722.
[19] LIU Y C，SHAO Z R，TENG Y Y，et al.NAM：normalization-based attention module[EB/OL].[2021-01-30].https://arxiv.org/abs/2111.12419.
[20] LIU L，FENG X F，LI F，et al.Surface defect detection of industrial components based on improved YOLOv5s[J].IEEE Sensors Journal，2024，24（15）：23940-23950.
[21] ZHOU X，ZHOU S H，ZHANG Y C，et al.GDALR：global dual attention and local representations in Transformer for surface defect detection[J].Measurement，2024，229：114398.
[22] LU W，CHEN S B，TANG J，et al.A robust feature downsampling module for remote-sensing visual tasks[J].IEEE Transactions on Geoscience and Remote Sensing，2023，61：1-12.
[23] WANG A，CHEN H，LIU L H，et al.YOLOv10：real-time end-to-end object detection[C]//Advances in Neural Information Processing Systems 37，2024：107984-108011.
[24] HAN K，WANG Y H，TIAN Q，et al.GhostNet：more features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway：IEEE，2020：1580-1589.
[25] YANG L X，ZHANG R Y，LI L D，et al.SimAM：a simple，parameter?free attention module for convolutional neural networks[C]//Proceedings of the International Conference on Machine Learning，2021：11863-11874.
[26] ZHENG Z H，WANG P，REN D W.Enhancing geo-metric factors in model learning and inference for o-bject detection and instance segmentation[J].IEEE Transactions on Cybernetics，2022，52（8）：8574-8586.
[27] ZHANG Y F，REN W Q，ZHANG Z，et al.Focal and efficient IOU loss for accurate bounding box regression[J].Neurocomputing，2022，506：146-157.
[28] HE Y，SONG K C，MENG Q G，et al.An end-to-end steel surface defect detection approach via fusing multiple hierarchical features[J].IEEE Transactions on Instrumentation and Measurement，2019，69（4）：1493-1504.
[29] REDMON J，FARHADI A.YOLOv3：an incremental improvement[EB/OL].[2018-04-18].https://arxiv.org/abs/1804.02767.
[30] ZHANG Y，GUO Z Y，WU J Q，et al.Real-time vehicle detection based on improved yolov5[J].Sustainability，2022，14（19）：12274.
[31] WANG C Y，BOCHKOVSKIY A，LIAO H Y M.YOLOv7：trainable bag-of-freebies sets new state-of the-art for real-time object detectors[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway：IEEE，2023：7464-7475.
[32] SOHAN M，RAM S T，REDDY R C V.A review on YOLOv8 and its advancements[C]//Proceedings of the International Conference on Data Intelligence and Cognitive Informatics，2024：529-545.
[33] REN S，HE K，GIRSHICK R，et al.Faster R-CNN：towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2015，39（6）：1137-1149.
[34] CAI X H，LAI Q X，WANG Y W，et al.Poly kernel inception network for remote sensing detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway：IEEE，2024：27706-27716.
[35] TANG S L，HE F，HUANG X L，et al.Online PCBdefect detector on a new PCB defect dataset[EB/OL].[2019-02-12].https://arxiv.org/abs/1902.06197.