Problem in converting digit image to numpy array

Question

I am practicing with the MNIST digit dataset. I was excited to make any method for converting my hand written digits to numpy array. But was not getting the expected result. Here's what I have written in photoshop:

Then I have written a code that convert the png image to np array that looks like this:

from PIL import Image
img  =Image.open('New Project (11).png')
img_arr = np.array(img)
img_arr = img_arr.flatten()
img_arr = img_arr.astype('float64')
some_digit = img_arr.reshape(28,28)

import matplotlib as mpl
import matplotlib.pyplot as plt

plt.imshow(some_digit,cmap='binary')

The result given was like this:

What wrong have I done or what is other easy alternative. Thanks for any help.

Answer 1

I found the problem. The numpy array that I have converted from the image is actually correct. For some reason it shows different. I have used skimage instead of PIL and inverted the image and it worked. The code is as follows:

import skimage.io as lo
from skimage import util 
img = lo.imread("New Project (16).png")
img = util.invert(img)
img_arr = np.array(img)
img_arr = img_arr.flatten()

# type(img_arr)
img_arr = img_arr.astype('float64')

some_digit = img_arr.reshape(28,28)
# some_digit = np.invert(some_digit)
# some_digit

import matplotlib as mpl
import matplotlib.pyplot as plt

# some_digit = X[0]
# some_digit_image = some_digit.reshape(28,28)

plt.imshow(some_digit,cmap='binary')