This program is used to detect drowsiness for any given person. In this program we check how long a person's eyes have been closed for. If the eyes have been closed for a long period i.e. beyond a certain threshold value, the program will alert the user by playing an alarm sound.

The program contains 3 files, which are

Files

• face_and_eye_detector_single_image.py - Detects face and eye from a single image. Demo-

• face_and_eye_detector_webcam_video.py - Detects face and eye in a webcam feed by user
• drowsiness_detect.py- This script detects if person is drowsy or not using webcam video feed

DEMO

Requirements

IMPORTANT

Download shape_predictor_68_face_landmarks.dat.bz2 from Shape Predictor 68 features Extract the file in the project folder using bzip2 -dk shape_predictor_68_face_landmarks.dat.bz2

numpy==1.15.2
dlib==19.16.0
pygame==1.9.4
imutils==0.5.1
opencv_python==3.4.3.18
scipy==1.1.0

Use pip install -r requirements.txtto install the given requirements.

Usage

Detect Face and Eyes in a Single Image

Put your file to be detected in images folder with name test.jpeg or change the file path in Line : 14 face_and_eye_detector_single_image.py to your image file.
Run script using:

python face_and_eye_detector_single_image.py

Detect Face and Eyes in a Webcam Feed

Run script using:

python face_and_eye_detector_webcam_video.py

Drowsiness Detection

Run script using:

python drowsiness_detect.py

The algorithm for Eye Aspect Ratio was taken from pyimagesearch.com blog, by Adrian RoseBrock.