Smart Gesture Interaction Enhancing Accessibility For Visually Impaired Individuals

The majority of visually impaired user assistance systems concentrate on particular activities, such as reading text or identifying obstacles, and frequently need users to switch between programs for different purposes. This work presents a mobile system that uses an OpenCV-based algorithm for real-time gesture recording and processing, allowing users to do a variety of help tasks using simple hand gestures. In the first step, a multi-head neural network architecture recognizes movements; in the second stage, it initiates activities such as environment recognition or object description. With 40,000 different images including synthetic gestures trained on the dataset, the system performs competitively in gesture classification and localization. For visually challenged users, this integrated method simplifies device control by providing an effective, multipurpose helper.