FAV experts work on translating Saudi sign language into text

Cooperation Science and research

Translating sign language from visual form into text has long been a primary research goal of Marek Hrúz's team at the FAV UWB. He has now established a collaboration with the Saudi Data & AI Authority, where they are working with them to release a dataset of Saudi sign language.

Until now, they have been working on sign language feature recognition, mainly in English. However, the portfolio of languages they can translate into text for people who do not know the sign language will likely expand soon. Marek Hrúz from the Department of Cybernetics at the Faculty of Applied Sciences (FAV) and his team attended a workshop at Johns Hopkins University last year, where they presented their sign language translation project to other colleagues and speech technology experts worldwide. Their work attracted the attention of representatives of the government institution Saudi Data & AI Authority (SDAIA) – one of the most critical institutions in Saudi Arabia – and they offered to have the Czech team continue their research in Saudi Arabia. During a fortnight-long workshop there, they developed a prototype system for translating American and Saudi sign language, focusing on multilingual translation.

Although the phrase Saudi Sign Language may seem unusual to readers, as Arabic, not Saudi, is the official language in Saudi Arabia, it is the correct term. "It is important to remember that sign languages are not a variant of spoken and written languages but are independent and natural languages originating in deaf communities. Sign language in Saudi Arabia has its grammar, its dialects and is not directly linked to spoken Arabic," explained Marek Hrúz.

"We exchanged data and information with our colleagues from Saudi Arabia on the issue of sign language translation, and together, we deployed the first Saudi sign language translation system. We established a connection with Maram Aljuaid, a renowned linguist working at King Faisal University, who has a close relationship with native speakers. We are working with her and her husband, Mohammad Alghannami, to publish a dataset of Saudi Sign Language for scientific purposes focusing on the linguistic component," explained Marek Hrúz.

Several other institutions are working on Saudi sign language and computer vision, but not at the level that the team around the FAV researcher has taken on. Therefore, from Marek Hrúz's point of view, the personal meeting was also beneficial for the Saudi experts. "We are going to develop cooperation in the data field - the creation of a new corpus of Saudi sign language, which could be an inspiration for Czech sign language and the deaf community in the Czech Republic. We also intend to develop a system for translating multilingual sign languages," concluded Marek Hrúz.

Gallery


Faculty of Applied Sciences

Martina Batková

06. 03. 2025