Demo Failure : Cannot detect face with funcntion face_cascade.detectMultiScale()

[Cool Brin]

This is the result:

This is the demo codes:

 

import numpy as np

import cv2

face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')

eye_cascade = cv2.CascadeClassifier('haarcascade_eye.xml')

img = cv2.imread('./nbaplayer/203500.png')

print img.shape

print img.size

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)

#print gray.shape

#print gray.size

#cv2.imshow('img1',gray)

cv2.imshow('hsv',hsv)

faces = face_cascade.detectMultiScale(gray,2,1)

print faces

for (x,y,w,h) in faces:

    img = cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)

    roi_gray = gray[y:y+h, x:x+w]

    roi_color = img[y:y+h, x:x+w]

    eyes = eye_cascade.detectMultiScale(roi_gray)

    for (ex,ey,ew,eh) in eyes:

        cv2.rectangle(roi_color,(ex,ey),(ex+ew,ey+eh),(0,255,0),2)

        

#cv2.rectangle(img,(38,0),(23,128),(255,255,255),3)

cv2.imshow('img',img)

cv2.waitKey(0)

cv2.destroyAllWindows()

 Where is wrong????

Thanks!!

[Michael Beyeler]

Hi,

If the pre-trained Haar classifier doesn't detect a face in your image, there are a few things you can try. The `detectMultiScale` function takes a few optional arguments that let you specify what to look for:

• `minFeatureSize`: The minimum face size to consider (for example, 20x20 pixels)
• `searchScaleFactor`: Amount by which to rescale the image (scale pyramid). For example, a value of 1.1 will gradually reduce the size of the input image by 10%, making it more likely for a face to be found than a larger value.
• `minNeighbors`: The number of neighbors each candidate rectangle should have to retain it. Typically, choose 3 or 5.

Then it is still possible for the eye detector to fail. It's a good idea to crop the image so that only the face region is retained, which is exactly what you're doing with `roi_gray`. Great! There are a few more things you can try:

• Use the same optional arguments as above to specify a typical eye size.
• Further crop `roi_gray` to remove for example the mouth region.
• Maximize the contrast of `roi_gray` using adaptive histogram equalization.
• Switch to a better eye cascade, such as `haarcascade_lefteye_2splits.xml` and `haarcascade_righteye_2splits.xml`. If some of the players are wearing glasses, use `haarcascade_eye_tree_eyeglasses.xml`.

More details can be found in Chapter 7 of the book.

Best,

Michael