CODEWITHSUNDEEP
python-3.xnumpytensorflow2.0
Arun
Arun

Reputation: 2478

Padding MNIST images from (28, 28, 1) to (32, 32, 1)

I am using the MNIST dataset from TensorFlow 2.0 and am trying to pad it with zeros and increase the image size from (28, 28, 1) to (32, 32, 1). The code is:

# Load MNIST dataset-
(X_train, y_train), (X_test, y_test) = tf.keras.datasets.mnist.load_data()

X_train.shape, y_train.shape
# ((60000, 28, 28), (60000,))

X_test.shape, y_test.shape
# ((10000, 28, 28), (10000,))

# Pad with 2 zeros on left and right hand sides-
X_train_padded = np.pad(X_train[:,], (2, 2), 'constant')

X_train_padded.shape
# (60004, 32, 32)

However, the 'np.pad()' function used above is not giving me the desired shape of (6000, 32, 32) and also, it returns arrays filled with zeros! Instead of the original values as in X_train.

Can you help?

I am using Python 3.8, TensorFlow 2.0 and numpy 1.18.

Thanks!

Upvotes: 2

Views: 2772

Answers (1)

stevemo
stevemo

Reputation: 1097

You're using numpy.pad wrong.

  1. Your array is (6000,32,32), so you only want to pad axis 1 and 2, not axis 0.
  2. The pad_width argument to np.pad works like this: ((axis 1 pad before, axis 1 pad after), ...) so if you want to pad 1 pixel on each side you should do ((0,0), (1,1), (1,1)). (Your code is padding all axes 2 on each side.)

Here's a toy example:

z = np.arange(12).reshape(3,2,2)
print(z.shape)
# (3, 2, 2)

z = np.pad(z, ((0,0),(1,1),(1,1)), 'constant')
print(z.shape)
# (3, 4, 4)

# as an example, take a look at the second "image"
print(z[1])
# [[0 0 0 0]
#  [0 4 5 0]
#  [0 6 7 0]
#  [0 0 0 0]]

Upvotes: 2

Related Questions