基于改进YOLOv7-tiny的带钢表面缺陷检测算法

doi:10.19678/j.issn.1000-3428.0068397

摘要/Abstract

摘要：

针对带钢表面缺陷检测任务存在的小目标检测效率低、缺陷定位不准确、检测算法参数量大、难以部署在终端设备上等问题, 提出一种改进的YOLOv7-tiny检测算法。首先, 使用GSConv替换颈部网络中的标准卷积, 基于GSConv设计一种改进的高效聚合网络(ELAN-G), 保证带钢表面缺陷特征信息被充分融合同时降低算法的参数量; 其次, 在预测头和颈部网络之间增加针对低分辨率和小缺陷的SPDConv模块, 模块生成一个中间特征图, 通过对中间特征图中的小缺陷特征信息进行过滤学习得到最终特征图, 以此提高预测头对小缺陷的检测精度; 最后, 引入MPDIoU损失函数, 合理利用边界回归框的几何性质, 简化损失函数计算过程并提高缺陷定位精度。实验结果表明, 在NEU-DET数据集上, 改进算法比其他6种先进目标检测算法效果更好, 性能更均衡, 其平均精度均值(mAP)可达74.1%, 且参数量和计算量低于所有对比算法, 可应用于工业环境中的带钢表面缺陷检测系统。

关键词: YOLOv7-tiny, 目标检测, 表面缺陷, GSConv, MPDIoU

Abstract:

An improved YOLOv7-tiny detection algorithm is proposed to address several challenges, including low efficiency in small-target detection, inaccurate defect localization, excessive parameters in the detection algorithm, and difficulties in deploying the model on terminal equipment for surface defect detection on strip steel. First, GSConv is introduced to replace the standard convolution in the Neck network, followed by the design of an improved and efficient aggregation network, ELAN-G, based on GSConv, which reduces the model parameters while ensuring adequate fusion of strip steel surface defect features. Second, the SPDConv module is integrated between the Head and Neck networks to improve detection of low-resolution and small defects. The module generates an intermediate feature map, which is subsequently filtered and processed to obtain the final feature map, improving the detection accuracy of the Head network for small defects. Finally, the MPDIoU loss function is adopted to leverage the geometric properties of the bounding regression box, simplifying the loss calculation process and enhancing defect localization accuracy. The experimental results indicate that the improved algorithm outperforms six other advanced target detection algorithms on the NEU-DET dataset, demonstrating a more balanced performance. The mean Average Precision (mAP) of the improved algorithm reaches 74.1%, while the parameter count and computational requirements are lower than those of all comparative algorithms, making it suitable for deployment in a steel surface defect detection system within an industrial environment.

Key words: YOLOv7-tiny, object detection, surface defect, GSConv, MPDIoU

阳丽莎, 李茂军, 胡建文, 王鼎湘. 基于改进YOLOv7-tiny的带钢表面缺陷检测算法[J]. 计算机工程, 2025, 51(1): 208-215.

YANG Lisha, LI Maojun, HU Jianwen, WANG Dingxiang. Strip Steel Surface Defect Detection Algorithm Based on Improved YOLOv7-tiny[J]. Computer Engineering, 2025, 51(1): 208-215.

https://www.ecice06.com/CN/Y2025/V51/I1/208

图/表 10

图1 改进YOLOv7-tiny算法的网络结构

Fig.1 Network structure of improved YOLOv7-tiny algorithm

图2 GSConv的原理

Fig.2 The principle of GSConv

图3 ELAN-G模块

Fig.3 ELAN-G module

图4 SPDConv的原理

Fig.4 The principle of SPDConv

图5 MPDIoU几何示意图

Fig.5 Geometric schematic diagram of MPDIoU

图6 NEU-DET数据集图片示例

Fig.6 Image examples of NEU-DET dataset

图7 2种算法的检测结果对比

Fig.7 Comparison of detection results between two algorithms

参考文献 25

1	李少波, 杨静, 王铮, 等. 缺陷检测技术的发展与应用研究综述. 自动化学报, 2020, 46 (11): 2319- 2336.
	LI S B , YANG J , WANG Z , et al. Review of development and application of defect detection technology. Acta Automatica Sinica, 2020, 46 (11): 2319- 2336.
2	周亮, 王振环, 孙东辰, 等. 现代精密测量技术现状及发展. 仪器仪表学报, 2017, 38 (8): 1869- 1878.
	ZHOU L , WANG Z H , SUN D C , et al. Present situation and development of modern precision measurement technology. Chinese Journal of Scientific Instrument, 2017, 38 (8): 1869- 1878.
3	PARK J K , KWON B K , PARK J H , et al. Machine learning-based imaging system for surface defect inspection. International Journal of Precision Engineering and Manufacturing-Green Technology, 2016, 3 (3): 303- 310. doi: 10.1007/s40684-016-0039-x
4	CHU M X , GONG R F , GAO S , et al. Steel surface defects recognition based on multi-type statistical features and enhanced twin support vector machine. Chemometrics and Intelligent Laboratory Systems, 2017, 171, 140- 150. doi: 10.1016/j.chemolab.2017.10.020
5	赵鹤, 杨晓洪, 李小彤, 等. 基于贝叶斯网络的铜带表面缺陷图像分类. 控制工程, 2022, 29 (10): 1901- 1906.
	ZHAO H , YANG X H , LI X T , et al. Image classification of copper strip surface defects based on Bayesian network. Control Engineering of China, 2022, 29 (10): 1901- 1906.
6	REN S , HE K , GIRSHICK R , et al. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39 (6): 1137- 1149. doi: 10.1109/TPAMI.2016.2577031
7	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. Washington D.C., USA: IEEE Press, 2023: 7464-7475.
8	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[EB/OL]. [2023-08-05]. https://arxiv.org/abs/1512.02325.
9	SHI X C, ZHOU S K, TAI Y C, et al. An improved Faster R-CNN for steel surface defect detection[C]//Proceedings of the 24th IEEE International Workshop on Multimedia Signal Processing. Washington D.C., USA: IEEE Press, 2022: 1-5.
10	LIU X, GAO J. Surface defect detection method of hot rolling strip based on improved SSD model[EB/OL]. [2023-08-05]. https://link.springer.com/chapter/10.1007/978-3-030-73216-5_15.
11	ZHONG H, WU B, ZHANG X, et al. Steel surface defect detection based on an improved YOLOv5 model[C]//Proceedings of the 5th International Conference on Intelligent Control, Measurement and Signal Processing. Washington D.C., USA: IEEE Press, 2023: 51-55.
12	曹义亲, 周一纬, 徐露. 基于E-YOLOX的实时金属表面缺陷检测算法. 图学学报, 2023, 44 (4): 677- 690.
	CAO Y Q , ZHOU Y W , XU L . A real-time metallic surface defect detection algorithm based on E-YOLOX. Journal of Graphics, 2023, 44 (4): 677- 690.
13	窦智, 胡晨光, 李庆华, 等. 改进YOLOv7的小样本钢板表面缺陷检测算法. 计算机工程与应用, 2023, 59 (23): 283- 292.
	DOU Z , HU C G , LI Q H , et al. Improved YOLOv7 algorithm for small sample steel plate surface defect detection. Computer Engineering and Applications, 2023, 59 (23): 283- 292.
14	HE K , ZHANG X , REN S , et al. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37 (9): 1904- 1916. doi: 10.1109/TPAMI.2015.2389824
15	LIN T Y, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE Press, 2017: 2117-2125.
16	LIU S, QI L, QIN H F, et al. Path aggregation network for instance segmentation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE Press, 2018: 8759-8768.
17	WANG C Y, YEH I H, LIAO H Y M. You only learn one representation: unified network for multiple tasks[EB/OL]. [2023-08-05]. http://arxiv.org/abs/2105.04206v1.
18	HOWARD A G, ZHU M L, CHEN B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications[EB/OL]. [2023-08-05]. http://arxiv.org/abs/1704.04861v1.
19	ZHANG X Y, ZHOU X Y, LIN M X, et al. ShuffleNet: an extremely efficient convolutional neural network for mobile devices[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D.C., USA: IEEE Press, 2018: 6848-6856.
20	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. Washington D.C., USA: IEEE Press, 2020: 1580-1589.
21	LI H L, LI J, WEI H B, et al. Slim-neck by GSConv: a better design paradigm of detector architectures for autonomous vehicles[EB/OL]. [2023-08-05]. http://export.arxiv.org/abs/2206.02424v2#:~:text=The%20GSConv%20accomplishes%20an%20excellent%20trade-off%20between%20the,robustly%20demonstrated%20in%20over%20twenty%20sets%20comparative%20experiments.
22	SUNKARA R, LUO T. No more strided convolutions or pooling: a new CNN building block for low-resolution images and small objects[EB/OL].[2023-08-05]. https://link.springer.com/chapter/10.1007/978-3-031-26409-2_27.
23	CARION N, MASSA F, SYNNAEVE G, et al. End-to-end object detection with transformers[EB/OL].[2023-08-05]. https://arxiv.org/abs/2005.12872.
24	ZHENG Z H , WANG P , LIU W , et al. Distance-IoU loss: faster and better learning for bounding box regression. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34 (7): 12993- 13000. doi: 10.1609/aaai.v34i07.6999
25	MA S L, XU Y, MA S L, et al. MPDIoU: a loss for efficient and accurate bounding box regression[EB/OL].[2023-08-05].http://arxiv.org/abs/2307.07662v1.

[1]	胡升龙, 陈彬, 张开华, 宋慧慧. 场景结构知识增强的协同显著性目标检测[J]. 计算机工程, 2025, 51(1): 31-41.
[2]	火久元, 苏泓瑞, 武泽宇, 王婷娟. 基于改进YOLOv8的道路交通小目标车辆检测算法[J]. 计算机工程, 2025, 51(1): 246-257.
[3]	李猛坤, 袁晨, 王琪, 赵冲, 陈景轩, 刘立峰. 基于改进YOLOv8算法的在线听课行为识别模型研究[J]. 计算机工程, 2025, 51(1): 287-294.
[4]	张天鹏, 韩晶, 吕学强. 基于多任务学习的超分辨率辅助小目标检测[J]. 计算机工程, 2024, 50(9): 304-312.
[5]	曾钰琦, 刘博, 钟柏昌, 钟瑾. 智慧教育下基于改进YOLOv8的学生课堂行为检测算法[J]. 计算机工程, 2024, 50(9): 344-355.
[6]	王昱婷, 刘志明, 万亚平, 朱涛. 基于可见光与红外图像的弱光条件下目标检测[J]. 计算机工程, 2024, 50(8): 270-281.
[7]	贵向泉, 刘世清, 李立, 秦庆松, 李唐艳. 基于改进YOLOv8的景区行人检测算法[J]. 计算机工程, 2024, 50(7): 342-351.
[8]	刘仕兵, 周诗涵. 高铁接触网绝缘子检测算法研究[J]. 计算机工程, 2024, 50(5): 200-208.
[9]	孙帮勇, 马铭, 于涛. 基于区域特征强化的多尺度伪装目标检测方法[J]. 计算机工程, 2024, 50(5): 209-219.
[10]	赵继达, 甄国涌, 储成群. 基于YOLOv8的无人机图像目标检测算法[J]. 计算机工程, 2024, 50(4): 113-120.
[11]	崔丽群, 曹华维. 基于改进YOLOv5的遥感图像目标检测[J]. 计算机工程, 2024, 50(4): 228-236.
[12]	周金涛, 高迪驹, 刘志全. 基于全景视觉的无人船水面障碍物检测方法[J]. 计算机工程, 2024, 50(2): 113-121.
[13]	任书玉, 汪晓丁, 林晖. 目标检测中注意力机制综述[J]. 计算机工程, 2024, 50(12): 16-32.
[14]	李沼洁, 朱恒亮, 毛国君, 杨鑫. 渐进式特征增强的弱监督显著性目标检测[J]. 计算机工程, 2024, 50(12): 233-244.
[15]	王林, 赵莉, 王无为. 高动态场景下无人机空对空目标检测[J]. 计算机工程, 2024, 50(12): 265-275.

选择文件类型/文献管理软件名称

选择包含的内容