Abstract
BACKGROUND: Artificial Intelligence (AI) offers transformative potential for human-computer interaction, particularly through eye-gesture recognition, enabling intuitive control for users and accessibility for individuals with physical impairments. METHODS: We developed an AI-driven eye-gesture recognition system using tools like OpenCV, MediaPipe, and PyAutoGUI to translate eye movements into commands. The system was trained on a dataset of 20,000 gestures from 100 diverse volunteers, representing various demographics, and tested under different conditions, including varying lighting and eyewear. RESULTS: The system achieved 99.63% accuracy in recognizing gestures, with slight reductions to 98.9% under reflective glasses. These results demonstrate its robustness and adaptability across scenarios, confirming its generalizability. CONCLUSIONS: This system advances AI-driven interaction by enhancing accessibility and unlocking applications in critical fields like military and rescue operations. Future work will validate the system using publicly available datasets to further strengthen its impact and usability.