Deep Learning-Based Automated Defect Detection in Solar Cell Images

Abstract

This research presents an automated deep learning-based methodology for detecting defects in solar cell images, emphasizing the practical application of the Xception architecture. Solar energy production relies heavily on the quality and efficiency of solar cells, making accurate defect detection critical for minimizing energy losses and production costs. Traditional manual inspection methods are time-consuming, inconsistent, and prone to human error, limiting scalability in industrial settings. By leveraging the depthwise separable convolutions and efficient architecture of Xception, the proposed system effectively extracts complex features from high-resolution solar cell images. This enables precise differentiation between defective and non-defective cells while maintaining computational efficiency, allowing the model to function effectively even in environments with limited hardware resources. A balanced and well-curated dataset of solar cell images was used to train and validate the Xception-based model. The dataset includes diverse defect types, capturing variations in cell surface anomalies, scratches, cracks, and other imperfections that impact performance. The model undergoes rigorous preprocessing, including image normalization, resizing, and data augmentation techniques, to improve generalization and reduce overfitting. Experimental results demonstrate that the Xception model achieves high classification accuracy while maintaining a lightweight footprint suitable for deployment in industrial scenarios. Performance metrics, such as precision, recall, and F1-score, confirm the model’s reliability in identifying defective cells, highlighting its potential to enhance automated quality control in solar panel manufacturing.