Published Paper Details:

Accurate classification of petrological samples is fundamental to geological investigations and resource exploration. Traditional methods relying on manual lithographic analysis are time-consuming and subject to expert bias. This study presents a novel deep learning framework for automated classification of Lithographical thin-section images. Utilizing convolutional neural networks (CNNs), the model was trained on a curated dataset encompassing diverse rock types, including igneous, sedimentary, and metamorphic classes. Data augmentation and transfer learning techniques enhanced model generalizability and performance. The proposed system achieved an overall classification accuracy exceeding 90%, outperforming conventional machine learning baselines. The results highlight the potential of deep learning to streamline petrological workflows, reduce human error, and enable large-scale geological mapping. Future work aims to integrate geochemical data and expand the model's applicability to field-acquired samples