AI-powered Eye Blink Detection for Health & Safety

Inspiration

The project was inspired by the growing issue of Digital Eye Syndrome (DES), caused by long hours of screen exposure. People blink significantly less while working on computers or mobile devices, leading to dryness, fatigue, and eye strain. We wanted to create a system that could monitor blink frequency in real time and provide useful insights for health and safety.

What it does

The system uses AI-powered eye blink detection to track and analyze a user’s blink rate. It helps identify irregularities in blinking, warns against prolonged eye strain, and can also be adapted for safety applications such as monitoring driver drowsiness.

How we built it

Used Mediapipe for real-time facial landmark detection. Implemented CNN + RNN models to classify eye states (open, closed, blink). Built a Python-based application capable of real-time video analysis. Conducted tests under different lighting conditions and with multiple subjects.

Challenges we ran into

Maintaining accuracy in varying lighting and backgrounds. Ensuring reliable detection when users wear glasses or masks. Collecting a diverse dataset to cover multiple scenarios. Optimizing the system for real-time performance without heavy computation.

Accomplishments that we're proud of

Successfully achieved real-time blink detection with high accuracy. Demonstrated the potential of the system in both health monitoring and safety applications. Built a lightweight solution that can run on personal devices without requiring specialized hardware.

What we learned

The importance of blink frequency in identifying eye fatigue and health issues. Techniques to integrate computer vision with AI models for practical applications. How to balance performance, accuracy, and real-time processing in applied AI systems.

What's next for AI-powered Eye Blink Detection for Health & Safety

Extending the system to a mobile app for wider accessibility. Integrating with IoT devices or smart glasses for continuous monitoring. Expanding applications into driver assistance systems to reduce accidents. Using blink data for early diagnosis of digital eye syndrome and related conditions.

Built With

• cnn
• deeplearning
• mediapipe
• python