Defect Detection of Industrial Aluminum Sheets Based on the Yolo Model

Yongtao Xu

doi:10.54097/v7tzq516

Authors

Yongtao Xu

DOI:

https://doi.org/10.54097/v7tzq516

Keywords:

Industrial aluminum sheet; Surface defect detection; YOLOv4; Lightweight network; Object detection.

Abstract

With the continuous improvement of industrial automation, surface quality inspection has become increasingly important in the manufacturing industry. To tackle the difficulties inherent in balancing precision and efficiency, as well as the large model size in industrial aluminum sheet defect detection, this paper puts forward a lightweight, high-precision improved detection method based on the YOLOv4 framework. The proposed method replaces the CSPDarknet53 backbone with GhostNet to reduce computation and parameter size; incorporates depthwise separable convolutions into the feature fusion layer; and replaces the SPP module with the SPPF module to enhance the efficiency of multi-scale feature fusion while accelerating inference speed. A channel attention mechanism is embedded in key feature layers to improve the detection of small-scale defects, and an IoU-based K-means clustering method is employed to optimize anchor generation, enhancing the matching degree between predicted boxes and ground-truth boxes. Experimental results show that the improved model outperforms the original YOLOv4 and other compared methods in terms of mAP, precision, recall, and detection speed on the industrial aluminum sheet defect dataset, while significantly reducing model size. This makes it suitable for meeting the dual requirements of real-time performance and detection accuracy in actual industrial production.

References

[1]Ren, S., He, K., Girshick, R., & Sun, J. (2017). Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6), 1137-1149.

[2]Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C. Y., & Berg, A. C. (2016). SSD:Single shot multibox detector. In Proceedings of the European Conference on Computer Vision. Springer, Cham, 21-37.

[3]Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 779-788.

[4]Redmon, J., & Farhadi, A. (2017). YOLO9000: Better, faster, stronger. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 7263-7271.

[5]Redmon, J., & Farhadi, A. (2018). YOLOv3: An incremental improvement. arXiv Preprint arXiv:1804.02767.

[6]Bochkovskiy, A., Wang, C. Y., & Liao, H. Y. M. (2020). YOLOv4: Optimal speed and accuracy of object detection. arXiv Preprint arXiv:2004.10934.

[7]Ding, R. W., Dai, L. H., Li, G. P., & Liu, Y. (2019). TDD-Net: A tiny defect detection network for printed circuit boards. CAAI Transactions on Intelligence Technology, 4(2), 110-116.

[8]Cheng, J. Y., Duan, X. H., & Zhu, W. (2021). Research on metal surface defect detection based on improved YOLOv3. Computer Engineering and Applications, 57(19), 252-258.

[9]Zhao, H. L., Yang, Z. F., & Li, J. (2021). Detection of metal surface defects based on YOLOv4 algorithm. Journal of Physics: Conference Series, 1907(1), 012043.

[10]Li, Q. D., & Li, T. L. (2021). Steel plate defect detection based on improved YOLOv3 algorithm. Electronic Measurement Techniques, 44(2), 104-108.

[11]Li, C. F., Cai, J. L., Qiu, S. H., & Zhang, X. Y. (2023). Method for detecting surface defects of aluminum materials based on improved YOLOv4. Chinese Test, 17.

[12]Liu, H., Chen, Z. L., & Wang, S. X. (2021). Tool surface defect detection based on improved YOLOv3 algorithm. Combined Machine Tools & Automated Machining Technology, (11), 87-90.