TensorFlow is open-source Python library designed by Google to develop Machine Learning models and deep learning neural networks.
Padding means adding values before and after Tensor values.
Method Used:
- tf.pad: This method accepts input tensor and padding tensor with other optional arguments and returns a Tensor with added padding and same type as input Tensor. Padding tensor is a Tensor with shape(n, 2).
Example 1: This example uses constant padding mode i.e. value at all the padded indices will be constant.
Python3
# importing the library import tensorflow as tf # Initializing the Input input = tf.constant([[1, 2], [3, 4]]) padding = tf.constant([[2, 2], [2, 2]]) # Printing the Input print("Input: ", input) print("Padding: ", padding) # Generating padded Tensor res = tf.pad(input, padding, mode ='CONSTANT') # Printing the resulting Tensors print("Res: ", res ) |
Output:
Input: tf.Tensor( [[1 2] [3 4]], shape=(2, 2), dtype=int32) Padding: tf.Tensor( [[2 2] [2 2]], shape=(2, 2), dtype=int32) Res: tf.Tensor( [[0 0 0 0 0 0] [0 0 0 0 0 0] [0 0 1 2 0 0] [0 0 3 4 0 0] [0 0 0 0 0 0] [0 0 0 0 0 0]], shape=(6, 6), dtype=int32)
Example 2: This example uses REFLECT padding mode. For this mode to work paddings[D, 0] and paddings[D, 1] must be less than or equal to tensor.dim_size(D) – 1.
Python3
# importing the library import tensorflow as tf # Initializing the Input input = tf.constant([[1, 2, 5], [3, 4, 6]]) padding = tf.constant([[1, 1], [2, 2]]) # Printing the Input print("Input: ", input) print("Padding: ", padding) # Generating padded Tensor res = tf.pad(input, padding, mode ='REFLECT') # Printing the resulting Tensors print("Res: ", res ) |
Output:
Input: tf.Tensor( [[1 2 5] [3 4 6]], shape=(2, 3), dtype=int32) Padding: tf.Tensor( [[1 1] [2 2]], shape=(2, 2), dtype=int32) Res: tf.Tensor( [[6 4 3 4 6 4 3] [5 2 1 2 5 2 1] [6 4 3 4 6 4 3] [5 2 1 2 5 2 1]], shape=(4, 7), dtype=int32)
