Python OpenCV GrabCut Foreground Detection

import	numpy  as  np
import	cv2
from	matplotlib  import pyplot  as	plt



img  =	cv2.imread('salah.jpg')
mask =	np.zeros(img.shape[:2],np.uint8)


bgdModel =  np.zeros((1,65),np.float64)
fgdModel =  np.zeros((1,65),np.float64)



rect =	(300,30,421,378)
cv2.grabCut(img,mask,rect,bgdModel,fgdModel,5,cv2.GC_INIT_WITH_RECT)


mask2  =  np.where((mask==2)|(mask==0),0,1).astype('uint8')
img  =	img*mask2[:,:,np.newaxis]


plt.subplot(121),  plt.imshow(img)
plt.title("grabcut"),	plt.xticks([]),	plt.yticks([])
plt.subplot(122),
plt.imshow(cv2.cvtColor(cv2.imread('salah.jpg'),
cv2.COLOR_BGR2RGB))
plt.title("original"),	plt.xticks([]),	plt.yticks([])


plt.show()

import numpy as np

import cv2

from matplotlib import pyplot as plt

img = cv2.imread('salah.jpg')

mask = np.zeros(img.shape[:2],np.uint8)

bgdModel = np.zeros((1,65),np.float64)

fgdModel = np.zeros((1,65),np.float64)

rect = (300,30,421,378)

cv2.grabCut(img,mask,rect,bgdModel,fgdModel,5,cv2.GC_INIT_WITH_RECT)

mask2 = np.where((mask==2)|(mask==0),0,1).astype('uint8')

img = img*mask2[:,:,np.newaxis]

plt.subplot(121), plt.imshow(img)

plt.title("grabcut"), plt.xticks([]), plt.yticks([])

plt.subplot(122),

plt.imshow(cv2.cvtColor(cv2.imread('salah.jpg'),

cv2.COLOR_BGR2RGB))

plt.title("original"), plt.xticks([]), plt.yticks([])

plt.show()

now in the above code we create zero-filled and foreground and background models

bgdModel   =	np.zeros((1,65),np.float64)
fgdModel   =	np.zeros((1,65),np.float64)

1 2	bgdModel = np.zeros((1,65),np.float64) fgdModel = np.zeros((1,65),np.float64)

Background and foreground models are going to be determined based on the areas left out of the initial rectangle

This is the rectangle code

rect =	(300,30,421,378)

1	rect = (300,30,421,378)

So now to the interesting part! We run the GrabCut algorithm specifying the empty models
and mask, and the fact that we’re going to use a rectangle to initialize the operation:

cv2.grabCut(img,mask,rect,bgdModel,fgdModel,5,cv2.GC_INIT_WITH_RECT)

1	cv2.grabCut(img,mask,rect,bgdModel,fgdModel,5,cv2.GC_INIT_WITH_RECT)

You’ll also notice an integer after fgdModel, which is the number of iterations the
algorithm is going to run on the image. You can increase these, but there is a point in
which pixel classifications will converge, and effectively, you’ll just be adding iterations
without obtaining any more improvements.
After this, our mask will have changed to contain values between 0 and 3. The values, 0
and 2, will be converted into zeros, and 1-3 into ones, and stored into mask2, which we can
then use to filter out all zero-value pixels (theoretically leaving all foreground pixels
intact):

mask2 =	np.where((mask==2)|(mask==0),0,1).astype('uint8')
img   =	img*mask2[:,:,np.newaxis]

1 2	mask2 = np.where((mask==2)\|(mask==0),0,1).astype('uint8') img = img*mask2[:,:,np.newaxis]

The last part of the code displays the images side by side

plt.subplot(121),	plt.imshow(img)
plt.title("grabcut"),	plt.xticks([]),	plt.yticks([])
plt.subplot(122),
plt.imshow(cv2.cvtColor(cv2.imread('salah.jpg'),
cv2.COLOR_BGR2RGB))
plt.title("original"),	plt.xticks([]),	plt.yticks([])


plt.show()