Publication

Abstract : Hand gestures serve as a natural and intuitive means for humans to communicate ideas and intents. For people with speaking or hearing disabilities, sign languages based on visual hand gestures are indispensable for their expression. With recent advances in machine perception and proliferation of Internet-of- Things (IoT) devices, automating gesture recognition to generate textual or speech output has transformative potential to augment accessibility. This paper surveys key innovations in gesture- to-text conversion spanning machine learning, deep learning and IoT-driven approaches. The research first motivate the assistive technology applications of an IoT framework leveraging specialized sensors and edge devices for gesture tracking, coupled with intelligent algorithms to interpret gestures and convert them into language. Our study covers convolutional and recurrent neural networks for spatial and temporal modelling, as well as multimodal fusion techniques. Challenges related to occlusion, sensor constraints and model generalization are discussed. The paper highlights the tremendous value of gesture-driven systems in enabling more immersive human-computer interaction and accessible interfaces for diverse communities.