https://doi.org/10.65770/QCNF4155
ABSTRACT
With the increasing global demand for renewable energy, the solar photovoltaic systems have been widely deployed. However, the operation efficiency and lifespan of solar panels are under the pressure of the surface defects. Defects such as dust and bird dropping has greatly affect the performance. Traditional inspection methods such as manual visual assessment are time-consuming and expensive. With the development of deep learning technology, the image classification techniques shows bright prospects in defect detection. Unfortunately, many existing models suffer from high computational complexity and limited interpretability, which restrict their practical application. This project proposes a lightweight deep learning-based image classification model for solar panel surface condition classification, aiming at solving these limitations. The proposed model integrates depthwise separable convolutions, residual Inception style branch structures and a custom attention mechanism to achieve efficient feature extraction ability with low computational cost. In addition, three complementary explainable artificial intelligence (XAI) techniques—Grad-CAM, LIME, and Occlusion Analysis are applied to enhance the model transparency and interpretability. The experimental results of four categories of solar panel image dataset demonstrate that the proposed model achieves a test accuracy of 75.62%, an F1-score of 75.97% and a ROC-AUC score of 0.9305, showing strong discrimination capability and good generalization performance.
References
- [1] P. Yadav, B. Sharma, and G. Olague, “Explainable attention-based fuzzy residual convolution network for solar cell defect identification and classification,” Eng. Appl. Artif. Intell., vol. 158, p. 111475, Oct. 2025, doi: 10.1016/j.engappai.2025.111475.
- [2] T. J. Kiran, “COMPUTER VISION ACCURACY ANALYSIS WITH DEEP LEARNING MODEL USING TENSORFLOW,” Int. J. Innov. Res. Comput. Sci. Technol., vol. 8, no. 4, July 2020, doi: 10.21276/ijircst.2020.8.4.13.
- [3] Ghaleb et al., “Explainable AI for Lightweight Network Traffic Classification Using Depthwise Separable Convolutions,” IEEE Open J. Comput. Soc., vol. 6, pp. 908–920, 2025,doi: 10.1109/OJCS.2025.3576495.
- [4] Tu, W. Liu, L. Zhao, and T. Yan, “Hyperspectral image classification based on deep separable residual attention network,” Infrared Phys. Technol., vol. 140, p. 105401, Aug. 2024, doi: 10.1016/j.infrared.2024.105401.
- [5] Dang, P. Pang, J. Lee, L. Dang, P. Pang, and J. Lee, “Depth-Wise Separable Convolution Neural Network with Residual Connection for Hyperspectral Image Classification,” Remote Sens., vol. 12, no. 20, Oct. 2020, doi: 10.3390/rs12203408.
- [6] Huanhuan, P. Guofeng, and Z. Hui, “Depthwise Separable Residual Network for Remote Sensing Image Scene Classification,” in 2022 International Symposium on Advances in Informatics, Electronics and Education (ISAIEE), Dec. 2022, pp. 533–539. doi: 10.1109/ISAIEE57420.2022.00115.
- [7] “Lightweight image matting algorithm based on deep learning – Qin – 2023 – IET Image Processing – Wiley Online Library.” Accessed: Dec. 15, 2025. [Online].
- [8] -H. Lee, L.-C. Yan, C.-S. Yang, S.-H. Lee, L.-C. Yan, and C.-S. Yang, “LIRNet: A Lightweight Inception Residual Convolutional Network for Solar Panel Defect Classification,” Energies, vol. 16, no. 5, Feb. 2023, doi: 10.3390/en16052112.
- [9] Chen, Z. Wang, and H. Li, “DS-ConvNeXt: A Lightweight Network for Visual Surface Defect Detection Based on Depthwise Separable Convolutions,” IEEE Transactions on Industrial Informatics, vol. 20, no. 3, pp. 3964-3973, Mar. 2024, doi: 10.1109/TII.2023.3321021.
- Sharma and P. Kumar, “A dual-attention convolutional neural network for automated defect detection and classification,” Expert Systems with Applications, vol. 216, p. 119459, May 2023, doi: 10.1016/j.eswa.2022.119459.
- Wang, F. R. C. Sousa, and M. H. Kim, “A quantitative evaluation framework for explainable AI in industrial defect diagnosis,” Engineering Applications of Artificial Intelligence, vol. 133, Part A, p. 107971, Jun. 2025, doi: 10.1016/j.engappai.2024.107971.
- Kumar, S. M. P. D. S. B. R. A. K. N. S. S. P. and G. B. , “Deep learning in photovoltaic fault detection: A review of techniques and future directions,” Solar Energy, vol. 262, pp. 111800, Sep. 2023, doi: 10.1016/j.solener.2023.111800.
- Zhang, K. Wei, and J. Li, “ECA-Net: Efficient Convolutional Attention for Deep Convolutional Neural Networks,” IEEE Access, vol. 12, pp. 4567-4578, Jan. 2024, doi:1109/ACCESS.2023.3349285.
Download all article in PDF
