SENSOR-EQUIPPED GLOVES FOR RECOGNIZING ARABIC SIGN LANGUAGE.

This study aims to propose a low-cost, smart glove-based sensory recognition system. This involves designing a glove containing multiple sensors to record data on static signs in Arabic Sign Language. The research was divided into several successive phases. The first phase included an analysis of the hand structure and movement in Arabic Sign Language to extract essential features that would help define the requirements of the recognition system. The study examined the language from several morphological and kinetic perspectives. The second phase focused on the design and development of the DataGlove through iterative refinement processes based on the requirements of the first phase, with the goal of improving its efficiency and performance. In the third phase, a sensory dataset was created based on the proposed glove, adopting a clear methodology for data collection and processing. The subsequent phase involved building a recognition system using machine learning algorithms, with a thorough evaluation of the effectiveness of the proposed methodology. The glove enabled the production of new, large-scale datasets for Arabic Sign Language (ARSL) using a sensory approach. The results demonstrated a number of significant benefits. The experiments demonstrated that the data glove design is an effective means of identifying similarities in hand postures and recording a wide variety of movements. The experiments achieved high accuracy, reaching 100%, in recognizing words, letters, and numbers using the Extra Trees algorithm. These results confirm that the proposed system achieves high recognition rates, and that the ARSL dataset will provide a supportive foundation for future research in Arabic sign language recognition systems. The results are encouraging, contributing to bridging the communication gap between the deaf and hard-of-hearing community and society.