Levelwise Degradation Classification of Ancient Stone Inscriptions Images using MobileNetV2: A Deep Learning Approach

This study presents a novel approach to classify ancient Karnataka stone inscriptions based on degradation levels using deep learning techniques. A custom dataset of 1,748 augmented images from 437 original images categorized into three degradation classes. The methodology employs MobileNetV2 architecture, benchmarked against CNN, ResNet50, and EfficientNet models. Through optimized preprocessing and augmentation techniques, the MobileNetV2-based model achieved 91% classification accuracy, significantly outperforming comparative models. This research offers an efficient, automated tool for assessing inscription conditions, contributing to digital archaeology and cultural heritage preservation. The study demonstrates the efficacy of lightweight deep learning models in historical artifact analysis and provides a foundation for future work in automated conservation strategies.