Vision Transformer-based Approach for Classification of Buddhist and Non-Buddhist Heritage Sites in Taxila

A Vision Transformer (ViT) model achieved 100% accuracy in classifying Buddhist and non-Buddhist heritage sites in Taxila, Pakistan, outperforming conventional convolutional neural networks like ResNet and EfficientNet. The study curated a diverse dataset of images from sites such as Takshasila University, Sirkap, Giri Fort, and the Dharmarajika Stupa, captured under varying conditions to ensure robustness. By leveraging ViT's ability to generalize across different visual features, researchers demonstrated its superior performance in preserving cultural heritage analysis. This approach offers a promising tool for efficient classification of ancient sites, advancing digital preservation efforts.

Abbas, M. S., Jamil, S., & Muhammad, A. (2025). Vision Transformer-based Approach for Classification of Buddhist and Non-Buddhist Heritage Sites in Taxila. 2025 International Conference on Emerging Technologies in Electronics, Computing, and Communication (ICETECC), 1–5. https://doi.org/10.1109/icetecc65365.2025.11070227