Driver Drowsiness Detection

Driver Drowsiness Detection

import cv2
import dlib
from scipy.spatial import distance
 
# Function to calculate the eye aspect ratio (EAR)
def eye_aspect_ratio(eye):
    A = distance.euclidean(eye[1], eye[5])
    B = distance.euclidean(eye[2], eye[4])
    C = distance.euclidean(eye[0], eye[3])
    ear = (A + B) / (2.0 * C)
    return ear
 
# Constants for drowsiness detection
EAR_THRESHOLD = 0.25  # Eye aspect ratio threshold for drowsiness detection
CONSECUTIVE_FRAMES = 20  # Number of consecutive frames for drowsiness detection
 
# Load face detector and facial landmark predictor
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor('path_to_shape_predictor_68_face_landmarks.dat')
 
# Load the video capture
video_capture = cv2.VideoCapture(0)  # Use 0 for webcam
 
# Initialize variables
frame_counter = 0
drowsy_frames = 0
alert = False
 
while True:
    ret, frame = video_capture.read()
    if not ret:
        break
 
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
 
    # Detect faces in the grayscale frame
    faces = detector(gray)
 
    for face in faces:
        shape = predictor(gray, face)
        shape = face_utils.shape_to_np(shape)
 
        # Extract eye coordinates
        left_eye = shape[36:42]
        right_eye = shape[42:48]
 
        # Calculate eye aspect ratios
        left_ear = eye_aspect_ratio(left_eye)
        right_ear = eye_aspect_ratio(right_eye)
 
        # Average the eye aspect ratios for both eyes
        avg_ear = (left_ear + right_ear) / 2.0
 
        # Check if the average eye aspect ratio is below the threshold
        if avg_ear < EAR_THRESHOLD:
            frame_counter += 1
            if frame_counter >= CONSECUTIVE_FRAMES:
                drowsy_frames += 1
                if not alert:
                    alert = True
                    print("Drowsiness detected!")
        else:
            frame_counter = 0
            alert = False
 
        # Draw the computed eye aspect ratio on the frame
        cv2.putText(frame, "EAR: {:.2f}".format(avg_ear), (10, 30),
                    cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 0, 255), 2)
 
    cv2.imshow("Driver Drowsiness Detection", frame)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
 
# Release the video capture and close the window
video_capture.release()
cv2.destroyAllWindows()

Note: Before running the code, make sure to download the "shape_predictor_68_face_landmarks.dat" file, which is the pre-trained model for facial landmark detection in dlib. You can download it from the following link: shape_predictor_68_face_landmarks.dat

Make sure to replace 'path_to_shape_predictor_68_face_landmarks.dat' with the actual path to the downloaded file on your system.

This code continuously captures video frames from a webcam, detects faces, and calculates the eye aspect ratio (EAR) to determine drowsiness. If the average EAR falls below a certain threshold for a specified number of consecutive frames, it triggers a drowsiness alert.

Feel free to modify and adapt the code to suit your specific requirements or integrate it into a larger system as needed.