科学技术与应用

基于Yolov11的金属表面缺陷检测算法研究

魏 妍妍

浙江师范大学

DOI: https://doi.org/10.59429/kxjsyy.v3i2.13836

Keywords: 金属表面缺陷检测；YOLOv11；坐标注意力；多分支聚合；微小目标检测；三维形貌

---

Abstract

针对保温杯三维形貌中凹坑、凸包等微小缺陷特征响应弱、易受背景噪声干扰的问题，提出一种基于改进 YOLOv11 的检测算法。构建多分支坐标注意力网络（MCA-Net），嵌入 YOLOv11 小尺度检测头，形成“局部增 强—全局建模”级联框架。实验表明，MCA-Net-YOLOv11 在 mAP@0.5 和 mAP@0.5:0.95 上分别达 0.752 和 0.756， 较基准提升 29.7% 和 83.9%，F1 值达 0.77。消融实验与可视化验证了模块协同增强效应，有效解决了复杂曲面下微 小三维缺陷检测难题。

---

References

[1] 李彦，侯庆彬，郑智等. 基于改进 YOLOv8 框架的钢材表面缺陷检测[J]. 模式识别，2025, 168: 111897.

[2] Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//European Conference on Computer Vision. Springer, 2020: 213-229.

[3] 孙毅，王磊，张华. 面向工业应用的轻量化高精度金属表面缺陷检测优化 YOLOv8[J]. 机器学习，2025, 114:226.

[4] Liu J G, Wang H Y. LFAF-DETR: Lightweight dynamic feature-sensing fusion network for industrial metal surface defect detection[J]. IEEE Transactions on Instrumentation and Measurement, 2025.

[5] 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,2023: 7464-7475.

[6] Jocher G, Chaurasia A, Qiu J. Ultralytics YOLO (Version 8.0)[R]. Technical Report, 2023.

[7] Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//European Conference on Computer Vision. Springer, 2020: 213-229.

[8] Terven J, Córdova-Esparza D M, Romero-González J A. A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and

YOLO-NAS[J]. Machine Learning and Knowledge Extraction,2023, 5(4): 1680-1716.

[9] Jocher G, Chaurasia A, Qiu J. Ultralyt-ics YOLO(Version 8.0)[R]. Technical Report, 2023.

[10] Hussain M. YOLO-v1 to YOLO-v8, the rise of YOLO and its complementary nature toward digital manufacturing and industrial defect detection[J]. Machines, 2023, 11(7):677.

[11] 李亮，董斌，赵敏. 基于改进元学习 YOLO 的钢管表面缺陷少样本目标检测模型[J]. 机电工程学报，2025,42(5): 985-993.

[12] Ye S, Wu J, Jin Y, et al. Novel variant transformer-based method for aluminum profile surface defect detection[J]. Measurement Science and Technology, 2025, 36(2):025602.

[13] 周杰，金勇，薛飞等. 基于改进 YOLOv11 的金属特种刀具表面微尺度缺陷检测模型构建[J]. 计量科学与技术，2026.

[14] Li Y, Yao T, Pan Y, et al. Contextual transformer networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 45(2): 1489-1500.

[15] 侯启迪，周东，冯建. 面向高效移动网络设计的坐标注意力机制[J]. 计算机视觉与模式识别，2021: 13713-13722.

[16] Ultralytics. YOLOv11 documentation[EB/OL].2024.

[17] 吴世荣，朴俊，李俊英等. 卷积块注意力模块[J].欧洲计算机视觉会议，2018: 3-19.

[18] Li Y, Yao T, Pan Y, et al. Contextual transformer networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 45(2): 1489-1500.