Smile detection

smileDetector.py

import cv2
import dlib


def faceLandmarks(im):

    # Path for the detection model, you can download it from here: https://github.com/italojs/facial-landmarks-recognition/blob/master/shape_predictor_68_face_landmarks.dat
    PREDICTOR_PATH = r"C:\Users\felipe.cunha\Documents\venv\cv2\week1-pyton\data\models\shape_predictor_68_face_landmarks.dat"
    
    # Create object to detect the face
    faceDetector = dlib.get_frontal_face_detector()

    # Create object to detect the facial landmarks
    landmarkDetector = dlib.shape_predictor(PREDICTOR_PATH)

    # Detect faces
    faceRects = faceDetector(im, 0)

    # Initialize landmarksAll array
    landmarksAll = []

    # For each face detected in the image, this chunk of code creates a ROI around the face and pass it as an argument to the 
    # facial landmark detector and append the result to the array landmarks 
    for i in range(0, len(faceRects)):
        newRect = dlib.rectangle(int(faceRects[i].left()),
                            int(faceRects[i].top()),
                            int(faceRects[i].right()),
                            int(faceRects[i].bottom()))
        landmarks = landmarkDetector(im, newRect)
        landmarksAll.append(landmarks)

    return landmarksAll, faceRects


def renderFacialLandmarks(im, landmarks):
    
    # Convert landmarks into iteratable array
    points = []
    [points.append((p.x, p.y)) for p in landmarks.parts()]

    # Loop through array and draw a circle for each landmark
    for p in points:
        cv2.circle(im, (int(p[0]),int(p[1])), 2, (255,0,0),-1)

    # Return image with facial landmarks 
    return im

# Read an image to a variable
im = cv2.imread("cv2/images/girlSmile.jpg")

# Get landmarks using the function created above
landmarks, _ = faceLandmarks(im)

# Render the landmarks on the first face detected. You can specify the face by passing the desired index to the landmarks array.
# In this case, one face was detected, so I'm passing landmarks[0] as the argument.
faceWithLandmarks = renderFacialLandmarks(im, landmarks[0])

# Calculate lips width
lips_width = abs(landmarks[0].parts()[49].x - landmarks[0].parts()[55].x)

# Calculate jaw width
jaw_width = abs(landmarks[0].parts()[3].x - landmarks[0].parts()[15].x)

# Calculate the ratio of lips and jaw widths
ratio = lips_width/jaw_width
print(ratio)

# Evaluate ratio
if ratio > 0.32 :
    result = "Smile"
else:
    result = "No Smile"

# Add result text to the image that will be displayed
cv2.putText(faceWithLandmarks, result, (50, 50), cv2.FONT_HERSHEY_COMPLEX, 1, (255, 0, 0), 2, cv2.LINE_AA)

# Display image
cv2.imshow("Smile Detection", faceWithLandmarks)
cv2.waitKey(0)