Cuneiform Reading Using Computer Vision Algorithms

Cuneiform Reading Using Computer Vision Algorithms

연구 분야: Artificial Intelligence

학회: SPML '22: Proceedings of the 2022 5th International Conference on Signal Processing and Machine Learning

This paper presents a new method for computer-assisted recognition of horizontal strokes in photographs of cuneiform tablets with 90,52 % accuracy. The cuneiform script is the oldest attested writing system in the world, used for over three thousand years throughout the ancient Near East, primarily by the cultures of Mesopotamia (modern Iraq). It was impressed on clay tablets and engraved on stone slabs by writing strokes. Researchers have been trying to speed up the process of reading the tablets using different methods, as manual copying of the tablets and their transliteration is time consuming. This research, therefore, aims to recognize the elementary components, i.e., the strokes, of cuneiform signs from photographs of ancient cuneiform tablets, in order to enable effective OCR using the latest computer vision algorithms. The main difference between other approaches and ours is that we work directly with the two-dimensional photographs, instead of three-dimensional models, as there are many more 2D images available in public online repositories. The goal is to partly automate the process of identifying and reading cuneiform signs, thus speeding up the process of rediscovering these ancient texts and civilizations.