Deep Learning-Based Defect Detection in Photovoltaic Panels
DOI:
https://doi.org/10.71222/9xfqn712Keywords:
defect detection, YOLOv8, NWD, Soft-NMSAbstract
The defect issues of photovoltaic solar panels are closely related to their efficiency and reliability. To enhance the efficiency and accuracy of defect detection in photovoltaic solar panels, this paper proposes an improved YOLOv8 model for this purpose. The improved C2f module, C2f-MS, is utilized to replace part of the original C2f structure in the model, which not only reduces the computational complexity and parameter count of the model but also enhances the extraction and fusion capabilities of multi-scale features. Additionally, NWD is incorporated into the existing CIOU to improve the detection performance for small targets, making the model's detection capabilities more balanced across various target sizes. Soft-NMS is employed to replace NMS, mitigating the issue of multiple detection boxes for a single target. Experimental results demonstrate that the improved YOLOv8 model achieves enhanced detection accuracy while reducing both the parameter count and computational complexity.
References
1. H. Sun, Q. Zhi, Y. Wang, Q. Yao, and J. Su, "China’s solar photovoltaic industry development: The status quo, problems and approaches," Appl. Energy, vol. 118, pp. 221–230, 2014, doi: 10.1016/j.apenergy.2013.12.032.
2. X. Wu, K. Yin, and Y. Lin, "Opportunities and challenges for developing distributed photovoltaic under the emission peak and carbon neutrality goal," in Proc. Int. Conf. Sustainable Technol. Manag. (ICSTM 2022), vol. 12299, pp. 25–34, Nov. 2022, doi: 10.1117/12.2644233.
3. J. Redmon, S. Divvala, R. Girshick and A. Farhadi, "You Only Look Once: Unified, Real-Time Object Detection," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 2016, pp. 779-788, doi: 10.1109/CVPR.2016.91.
4. A. G. Howard, M. Zhu, B. Chen, et al., "Mobilenets: Efficient convolutional neural networks for mobile vision applications," arXiv preprint arXiv:1704.04861v1, 2017.
5. W. Lin, Z. Wu, J. Chen, J. Huang and L. Jin, "Scale-Aware Modulation Meet Transformer," 2023 IEEE/CVF International Con-ference on Computer Vision (ICCV), Paris, France, 2023, pp. 5992-6003, doi: 10.1109/ICCV51070.2023.00553.
6. Yu, J., Jiang, Y., Wang, Z., et al., "Unitbox: An advanced object detection network," Proc. 24th ACM Int. Conf. Multimedia, 2016, pp. 516-520, doi: 10.1145/2964284.2967274.
7. J. Wang, C. Xu, W. Yang, et al., "A normalized Gaussian Wasserstein distance for tiny object detection," arXiv preprint arXiv:2110.13389v2, 2021.
8. Z. Zheng, P. Wang, W. Liu, et al., "Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression," Proc. AAAI Conf. Artif. Intell., vol. 34, no. 7, pp. 12993–13000, 2020, doi: 10.1609/aaai.v34i07.6999.
9. A. Neubeck and L. Van Gool, "Efficient Non-Maximum Suppression," 18th International Conference on Pattern Recognition (ICPR'06), Hong Kong, China, 2006, pp. 850-855, doi: 10.1109/ICPR.2006.479.
10. N. Bodla, B. Singh, R. Chellappa and L. S. Davis, "Soft-NMS — Improving Object Detection with One Line of Code," 2017 IEEE International Conference on Computer Vision (ICCV), Venice, Italy, 2017, pp. 5562-5570, doi: 10.1109/ICCV.2017.593.
11. J. Redmon and A. Farhadi, "YOLOv3: An Incremental Improvement," arXiv preprint arXiv:1804.02767v1, 201.
12. S. Qiao, L. -C. Chen and A. Yuille, "DetectoRS: Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 10208-10219, doi: 10.1109/CVPR46437.2021.01008.
13. C. -Y. Wang, A. Bochkovskiy and H. -Y. M. Liao, "YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors,"2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, BC, Canada, 2023, pp. 7464-7475, doi: 10.1109/CVPR52729.2023.00721.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Lele Liu (Author)
This work is licensed under a Creative Commons Attribution 4.0 International License.
