This study focuses on document image analysis and text analysis of Indonesian lontar manuscripts. It covers handwriting recognition, transliteration, and translation to make indigenous knowledge digitally accessible for research and education.
This project is dedicated to studying and digitizing lontar manuscripts — historical palm-leaf texts that have preserved centuries of Indonesian history, language, and culture. Our work brings together artificial intelligence, paleography, and heritage studies.
We aim to safeguard this invaluable cultural heritage by developing systems for handwriting recognition, transliteration, translation, and text reconstruction. These tools support historical Indonesian languages such as Old Sundanese, Old Javanese, Old Balinese, and others, helping to create digital archives that are accessible to both scholars and the wider public.
Lontar is a particular kind of writing material made of palm leaves used in Bali, Lombok, Java, Madura, and other islands in Indonesia. The leaves are carefully processed to create a surface suitable for writing. The letters were carved onto the surface using a small knife called pangutik or pangrupak. This carving technique creates shallow grooves inscribing the letters. The letters are nearly invisible. To make it more visible, burnt candlenut (kemiri) was applied to the surface, darkening the grooves and making the letters easier to read.
A manuscript consist of a bundle of palm leaves is also called lontar. Each sheet of lontar (lempiran) is tied together by a piece of cord. If you look closely at a lontar, you will notice a circular hole in the center, also in the right and left side. This hole allows the sheets to be tied together with a cord made from processed lontar fibers or any other materials.
Another important element is cakep, a protective cover similar to a book cover, sized to match the dimensions of the lontar sheets it protects. This cakep can be made of various materials, such as bamboo and wood. Palm leaves manuscript covered by cakep is also called cakepan. Some manuscripts are kept in wooden boxes called kropak. Sometimes the kropak is also decorated with carvings or paintings. A manuscript kept in a kropak is also called kropak or kropakan.
Because lontar manuscripts are centuries old, they require special care to ensure their preservation. Despite efforts like storing them in controlled environments and digitization, deterioration still occurs. Digital reconstruction of damaged manuscripts is an important way to ensure future generations can continue to learn from this heritage.
To develop tools for handwriting recognition, transliteration, translation, and text reconstruction that support the understanding and dissemination of historical lontar manuscripts.
We use image processing, image analysis, deep learning, and language modeling techniques to analyze both the visual and textual content of palm-leaf manuscripts. As part of our digitization techniques, we apply hyperspectral imaging to enhance the readability and recovery of faded or damaged text
Our interdisciplinary research is shared in the fields of AI, linguistics, history, and cultural preservation. Access our results through open-access platforms and academic journals.
Explore our published work across digital heritage, AI, linguistics, and Southeast Asian manuscript studies.
Paulus, Erick; Burie, Jean-Christophe; Verbeek, Fons J. · 2023
Paulus, Erick; Burie, Jean-Christophe; Verbeek, Fons J. · 2021
Yosephin A. Rahayu · 2019
We are here for those of you who are interested in combining technology with palm manuscripts from around the world.
Fons J. Verbeek
Leiden University
Erick Paulus
Leiden University
Universitas Padjadjaran
Willem van der Molen
Leiden University
Universitas Indonesia
Yosephin Apriastuti Rahayu
Leiden University
Universitas Gadjah Mada
Jean-Christophe Burie
La Rochelle University
Rahmat Sopian
Universitas Padjadjaran
Dian Nursantika
Leiden University
Universitas Terbuka
Setiawan Hadi
Universitas Padjadjaran
Email: f.j.verbeek {et} liacs.leidenuniv.nl
We welcome collaborations, inquiries, and feedback. Contact us to explore opportunities in digital humanities, language preservation, or historical document analysis.