Three Universitas Indonesia (UI) students created a sign language translation tool “TRANSAURA” using TensorFlow and Raspberry Pi technology. The Transaura team consisting of three students from the Universitas Indonesia (UI), namely Daffa Fairuzaufa Athallah Raharjo (UI Faculty of Engineering, 2020), Aine Shahnaz Tjandraatmadja (UI Faculty of Nursing, 2020), and Almaz Scarletta Tjakrashafanti (Faculty of Nursing, UI, 2020) UI Cultural Sciences, 2018).
They take great interest in sign language because they realize the lack of accessibility of sign language translation for the general public. “This raises social issues for persons with disabilities, such as educational disparities, inequality in employment opportunities, and the inclusion of social participation. Transaura technology itself is made using TensorFlow for machine learning and Raspberry Pi for object detection,” said Dr. Ir. Dodi Sudiana, M.Eng, FTUI Image Processing expert.
The Transaura team under the guidance of Dodi Sudiana, a lecturer at the Department of Electrical Engineering, Faculty of Engineering UI (FTUI), won third place at the national level in the grant competition for national research, the 2021 Tanoto Student Research Awards, in the field of appropriate technology. The team conducted research and underwent a gradual selection starting from the university level to the national level in June 2021 – January 2022. In the competition, the Transaura Team competed with 24 other teams. The Dean of FTUI, Prof. Dr. Heri Hermansyah, ST., M.Eng., IPU, hopes that this research can be continued to develop a more friendly environment for people with disabilities in Indonesia.
“Many deaf friends who graduate from Special Schools (SLB) have difficulty in the job search process or difficulty in accessing various public facilities. Our goal in developing Transaura is to make it easier for deaf friends to be able to communicate in two directions. The design of Transaura is in the form of a portable box that can be placed anywhere. This tool has two sides, the first side is for the deaf and the other side is for hearing friends,” said Daffa, the originator of Transaura’s idea.
“Two screens on the front and back allow for two-way communication. The first screen will be where the sign language interpreter uses object detection with the help of TensorFlow. The second screen will output the text located on the back side of the tool. The main component that is the brain of Transaura is the Raspberry Pi microprocessor,” said Almaz explaining the Transaura component.
TensorFlow is an open source library for large-scale numerical computing and machine learning. TensorFlow can train and run deep neural networks for handwritten digit classification, image recognition, word embedding, iterative neural networks, sequence-to-sequence models for machine translation, natural language processing, and PDE (partial differential equation) based simulations. Interestingly, TensorFlow supports large-scale production predictions with the same model that can be used for training.
Transaura is made to be used in office areas, supermarkets and transportation facilities. “With Transaura, it is hoped that equality for people with disabilities in various fields of work can be created, as the name implies Transaura (translating aura). Opportunity and equality for people with disabilities is the fulcrum of this research,” said the leader of the Transaura team, Aine.
***
Public Communication Bureau
Faculty of Engineering, Universitas Indonesia