Solar cell surface defects significantly impact energy efficiency, making accurate detection crucial before deployment. While traditional inspection methods struggle with low accuracy and limited defect classification, this project proposes an enhanced YOLOv5-based model combined with advanced image augmentation techniques—Mosaic, Mixup, HSV transformation, scaling, and flipping—to improve detection precision. The architecture further incorporates Coordinate Attention (CA) to enhance feature extraction and employs a decoupled detection head to resolve classification-localization conflicts. The model achieved a 97% accuracy rate on the ELPV dataset. As an extension, we trained the same dataset on YOLOv6, which outperformed all benchmarks with 99% accuracy, showcasing superior robustness and precision. Comparative evaluations with Faster R-CNN and optimized YOLOv5 across metrics like precision, recall, F-score, and mAP confirm the effectiveness of YOLOv6. Thus, the extended framework ensures faster and more reliable solar cell defect detection, supporting quality control in photovoltaic manufacturing.