Communicating with the Deaf and Hard of Hearing through Automatic Arabic Sign Language Translator

Abstract

Abstract The needs of persons with disabilities are still being undermined even though persons with disabilities have proven to be productive members of the community when given the right education, support, and tools. Assistive devices are being designed, developed, and sold to provide persons with disabilities the tools to be able to become productive members. With the advent of new technologies, software development has also concentrated on the needs of persons with disabilities. With the widespread of software’s especially on mobile devices, researchers can now develop tools for the community to communicate, work, collaborate and deal normally with persons with disabilities. In this paper, an automatic Artificial Intelligence (AI) based Arabic Sign Language Translator is proposed. The system is able to capture the picture of the sign language performed by a person who is deaf or hard of hearing and provide real-time translation of the hand gestures. Novel features are proposed to be extracted from the images to be input to the four different classifiers namely; Random Forest (RF), Bagging classifier, Regression, and Random Tree (RT). The result shows that the Random Forest classifier with proposed extract features performs better by achieving 92.15% accuracy as compared using input image directly with classifiers.