EyeSaver

Eye strain preventer that won most technically advanced award in the IEEE Winter 2021 Quarterly Project at UC San Diego. It uses four neural nets and a binary classifier to track your gaze as you use a computer and alert you when you’ve been looking at a screen for too long so you can take a break.

Requires:

• Hardware:
  • OpenCV AIKit camera module
• Infrastructure:
  • MQTT server running on port 30300 (can be reconfigured)
• Python Libraries:
  • cv2
  • depthai
  • paho-mqtt
  • pandas
  • sklearn

Steps to use:

1. $ python3 calibrate_eyes_ON.py -c. When ready, press enter and look around at your computer screen until the program finishes (about 15 seconds).
2. $ python3 calibrate_eyes_OFF.py -c. When ready, press enter and look around away from your computer screen until the program finishes (about 15 seconds).
3. $ python3 train_model.py. Train the ON/OFF screen binary classifier on the data you just generated.
4. $ python3 eye-strain-alerter.py. Start the alert system software.
5. $ python3 eye-monitor.py -c. Start the camera stream that monitors your gaze using the binary classifier.