COMPUTER VISION SYSTEMS FOR DOCUMENTING, PRESERVING, AND ANALYZING TRADITIONAL VISUAL ART FORMS

Abstract [English]

Traditional visual art forms are highly essential in preserving culture but conventional forms of documentation usually have shortcomings with regards to scalability, accuracy and durability of access by the user. In the present paper, the author provides a detailed computer vision-oriented system that is intended to document, preserve and analyze traditional artworks through the use of modern imaging and deep learning algorithms. The offered system incorporates multi-phase pipeline that includes the steps of image capture, image processing, semantics, and systematic storage that provides the opportunity to archive high-fidelity digital images. The framework can be applied to perform automated functions, e.g., style classification, damage detection, and authenticity assessment with a high degree of precision, with the help of convolutional neural networks (CNNs) and Vision Transformers (ViTs). This is suggested to have an efficient data collection procedure and annotation software incorporating images in museums, historical records and field surveys and a unified labeling process to be identical. In comparison to the traditional ways of documentation, the experimental data is shown to be highly accurate, feature extraction quality and efficient analysis. Besides, the system helps to use applications in digital restoration, cultural analytics and heritage conservation planning. The proposed framework would have given a high-scalable, intelligent, and reliable infrastructure of securing the traditional visual art in the digital era and allow additional computational analysis to the researchers and conservationists.