Frequently Asked Questions

BlinkEase is an intelligent, seamless companion designed to empower users to take control of their eye health by addressing digital eye strain caused by prolonged screen use.

BlinkEase aims to solve the widespread problem of digital eye strain, which manifests as symptoms like dry, tired, or itchy eyes, blurred vision, and headaches, often due to a decreased blink rate during screen focus.

BlinkEase helps by tracking key metrics like blinks per minute (BPM), duration of eye rest, and frequency of breaks, providing data-driven insights and fostering mindful awareness of blinking and taking breaks.

BlinkEase uses your computer's webcam. It first detects a face using a YuNet AI model, then performs facial landmark detection to identify 12 key points around the eyes. From these points, it calculates the Eye Aspect Ratio (EAR) to determine eye openness and count blinks.

The Eye Aspect Ratio (EAR) is a single numerical score calculated from the coordinates of eye landmarks that represents how open a user's eyes are. A high EAR indicates open eyes, while a drop towards zero indicates a blink. The system registers an eye as "closed" when the EAR falls below a default threshold (0.2).

Yes, BlinkEase uses AI. It employs a YuNet AI model for precise face detection and a second, more detailed AI model for facial landmark detection, which identifies 478 points on the face to understand eye contours.

BlinkEase classifies blinks by their duration: Partial (less than 80ms), Brief (less than 100ms), Typical (100-400ms), Delayed (400-700ms), Prolonged (700-1000ms), and Microsleep (1 second or longer). This classification provides insight into the quality of eye rest.

If the YuNet model can no longer detect a face with high confidence, the eye state is classified as "INVALID". The application intelligently concludes that the user has taken a break from the screen when the eye state remains "INVALID" for a continuous period.

The "INVALID" eye state occurs when the face detection model can no longer confidently identify a user's face. This state is crucial for accurately logging breaks, as it indicates the user might have turned away from the screen or left their desk.

No. BlinkEase is deeply committed to user privacy. All camera feed processing happens live, on your device, and is never stored or transmitted anywhere.

BlinkEase processes all camera data locally on your device. No video or image data leaves your computer, ensuring your privacy is maintained.

The "20-20-20 rule" recommends that for every 20 minutes of screen time, you take a 20-second break to look at something 20 feet away. BlinkEase tracks the frequency and duration of your breaks, helping you adhere to this rule for better eye health.

BlinkEase compiles all collected data into a Session Summary. Users can view this data in easy-to-read charts, track their habits over time, and see how they are progressing toward their wellness goals.

For optimal eye health, you should aim for a good blinking rate (above your set threshold) and a high fraction of Typical Blinks. Incomplete blinks (Partial or Brief) do not effectively combat dry eyes, while signs of fatigue (Delayed, Prolonged, or Microsleep blinks) suggest the need for rest.

Yes, the threshold for the Eye Aspect Ratio (EAR) that determines a "closed" eye state is adjustable in the settings to fine-tune the detection for each user's unique features.

For a deeper dive into the implementation, you can refer to the two-part blog series linked in the white paper: "Eye State Detection in ML.NET (Part 1)" and "Eye State Detection in ML.NET (Part 2)".