The project, which runs on a NVIDIA Jetson Nano 2GB Developer Kit, monitors the eyes of the user and voices a prompt when their blink rate is less than the recommended rate of 10 blinks per minute.
Thirteen-year-old Adrit Rao, was awarded the Jetson Project of the Month for his Blink Detection and Reminder (Blinkr). The project, which runs on a NVIDIA Jetson Nano 2GB Developer Kit, monitors the eyes of the user and voices a prompt when their blink rate is less than the recommended rate of 10 blinks per minute.
Several studies have shown that low eye blinking rate, usually triggered by the use of a computer screen, is the leading cause of computer vision syndrome and other related disorders. To address this problem, Adrit created Blinkr with a simple setup of Jetson Nano 2GB Developer Kit, a webcam (or a Raspberry Pi v2 camera), a speaker and a few other basic peripherals.
The camera monitors the face of the user and feeds the frames to the Jetson Nano. To detect blinking, Adrit uses a 68 point facial landmark pre-trained model available in the Dlib open source library. Eyes are detected in each frame and the eye aspect ratio (EAR) is calculated and used to record the number of blinks over time. When the total blinks in a minute is less than the recommended rate, the speaker voices an alarm urging the user to blink more.
Many of us working from home do not have the usual prompts or interruptions during our day to move away from our screens. Tools like Blinkr can help us adopt healthy screen habits. This is a great project to build at home to learn about Jetson and AI, and to protect your eyesight.
This project earned Adrit his Jetson AI Specialist certificate. We are keeping our appreciative (and healthy) eyes peeled out to see what he builds next. If you’re interested in building your own Blinkr, he has shared the instructions and the code here.