AMPITET

The AMPITET —Spanish acronym for Aproximación multidisciplinar al Procesamiento de imagen y lenguaje en los Textos
del Antiguo Egipto usando Transformers— project (2025) aims to design image processing systems for Ancient Egyptian texts for study, transliteration, and translation into modern languages. Specifically, it investigates techniques for converting the visual style of text images obtained from photographs of coffins, papyri, or walls into a standardized style that facilitates text study and the application of character recognition systems developed by the project's research team. In addition, AMPITET investigates natural language processing techniques using Transformer neural networks for converting transliterations into their respective English translations and their semantic analysis at the word, sentence, or paragraph level. The project is supported by data and methods developed in the research team's previous projects, OCR-PT-CT (2022) and TTAE (2024) projects. In the case of the former, transcription algorithms were designed from facsimile text images, while in the case of the latter, systems were developed to obtain text transliterations from their transcriptions. Thanks to these techniques, sufficiently large datasets have been built to train the learning systems proposed for this project.