Shilpasmriti: Digital Preservation of Ancient Temples

Ancient temple inscriptions constitute invaluable historical records that document linguistic evolution, dynastic history, socio-cultural practices, and religious traditions. However, continuous exposure to environmental degradation, biological growth, and human interference has rendered many inscriptions fragile and increasingly illegible. Manual documentation methods are time-intensive and require specialized epigraphic expertise, limiting large-scale preservation efforts. This paper presents ShilpaSmriti, an AI-driven digital preservation framework designed to automatically capture, interpret, translate, and archive ancient South Indian temple inscriptions. The proposed system integrates image preprocessing, script identification, optical character recognition (OCR), neural machine translation, and cloud-based storage to convert inscription images into searchable, multilingual digital records. Scripts such as Tamil-Brahmi, Grantha, Old Kannada, and early Telugu are processed using script-specific OCR pipelines. Experimental evaluation demonstrates effective text extraction, reliable translation, and secure long-term preservation through cloud infrastructure. The framework enables global accessibility to inscription data while safeguarding cultural heritage from physical deterioration