AI Based Eye Communication System for Paralyzed People
Pooja Sahu (0205poojasahu@gmail.com)
Ms. Priyanka Bande (Bandepinka8839@gmail.com)
Shri Rawatpura Sarkar University, Raipur (C.G.)
Abstract:
Paralyzed patients often face severe communication barriers, relying entirely on external assistance to convey even the most basic needs. Existing assistive systems such as EEG-based Brain–Computer Interfaces (BCIs) are either prohibitively expensive or technically complex, making them inaccessible to most patients. This paper presents an AI-Based Eye Communication System, designed as a low-cost, non-invasive alternative that enables paralyzed individuals to communicate using only eye movements and blinks. The proposed model integrates computer vision and machine learning techniques through standard HD cameras to detect ocular gestures such as left, right, and upward gaze in real time. These gestures are mapped to predefined commands like “Yes,” “No,” and “Help,” with outputs delivered via synthesized voice or automated caregiver alerts through WhatsApp and email. Developed using OpenCV and MediaPipe frameworks, under standard lighting conditions. Additionally, an integrated emergency-detection module identifies rapid blink sequences to trigger urgent alerts automatically. Experimental results show over 95% accuracy with response times under 220 ms, confirming its real-time feasibility. The modular design of the system ensures easy customization for different users and environments, while its portable architecture allows deployment in hospitals, rehabilitation centers, and home-based care. Beyond healthcare, the proposed system can be extended toward AI-driven smart environments, where eye gestures could control assistive devices and IoT applications. This research thus marks a step forward in democratizing AI-powered assistive communication, bridging the gap between technological innovation and social impact.
Keywords:
Eye tracking, gesture recognition, paralyzed patients, computer vision, assistive technology, OpenCV, MediaPipe, AI communication,human–computer interaction, real-time processing, emergency detection.
