Abstract
Total communication in sign language is a communication approach that involves several components in conveying information, including hand movements, body movements, lip movements, and facial expressions. Currently, there is no publicly available Indonesian Sign Language (SIBI) dataset. Therefore, researchers must collect their own data. Utilizing diverse datasets can result in partiality when making inferences about the suggested approach. The SIBI dataset was built by involving 10 teachers and 10 hearing-impaired high school students. The SIBI dataset includes basic signs and sign sentences. Basic sign gestures include letters, numbers, and affixes, according to the SIBI online dictionary published by the Indonesian Government. Sentences are created by several teachers using common communication phrases during learning process. Sign language sentences are delivered using total communication, which combines lip movements, body movements, and facial expressions. The same sentence can be delivered with different facial expressions, giving different meanings. This article discusses how a sign sentence is expressed with different facial expressions. In SIBI, word morphology follows the rules of the Indonesian language. Therefore, this paper also explains the morphemic structure of the 40 sentences used as references. The use of the SIBI dataset can be adjusted to certain research parameters. The video dataset can be converted into a series of frames or images as needed for the research. Although audio is not available, the SIBI video data captures lip movements that can be used in lip reading research. The SIBI dataset can facilitate research in computer vision as well as other fields of study related to sign language.