Tensorflow.js tf.mirrorPad() Function
Last Updated :
21 Jun, 2021
Tensorflow.js is an open-source library that is developed by Google for running machine learning models as well as deep learning neural networks in the browser or node environment.
The .mirrorPad() function is used to pad the stated tensor input with the help of mirror padding. Moreover, this method is beneficial in implementing the REFLECT as well as SYMMETRIC modes of pad.
Syntax :
tf.mirrorPad(x, paddings, mode)
Parameters:
- x: It is the stated tensor which is to be padded, and it can be of type tf.Tensor, TypedArray, or Array.
- paddings: It is an array whose length is R, that is the order of the stated tensor. Where, all the elements form a tuple of length two i.e. ints [padBefore, padAfter] that specifies to what extent it is to be padded with every size of the stated tensor. Moreover, in the reflect mode of padding, the padded sections should exclude the extremities, whereas in the symmetric mode of padding the padded sections should not exclude the extremities. And it is of type array.
- mode: It specifies the mode of padding i.e. either reflect or symmetric. And it is of type string.
Note:
- Firstly, if the stated tensor input is [4, 5, 6] and paddings is [0, 1], then the output will be [4, 5, 6, 5] in the reflect mode of padding and [4, 5, 6, 6] in the symmetric mode of padding.
- Secondly, if the mode of padding is reflect then the paddings[D, 0] as well as paddings[D, 1] must not be higher than x.shape[D] – 1, whereas if the mode of padding is symmetric then the paddings[D, 0] as well as paddings[D, 1] must not be higher than x.shape[D].
Return Value: It returns tf.Tensor object.
Example 1:
Javascript
import * as tf from "@tensorflow/tfjs"
const y = tf.tensor1d([4, 5, 6]);
const padding = [[0, 1]];
const mode = 'reflect' ;
var res = tf.mirrorPad(y, padding, mode);
res.print();
|
Output:
Tensor
[4, 5, 6, 5]
Example 2:
Javascript
import * as tf from "@tensorflow/tfjs"
tf.mirrorPad(tf.tensor(
[2.4, 6.8, 9.3, 5.3]),
[[0, 2]], 'symmetric' ).print();
|
Output:
Tensor
[2.4000001, 6.8000002, 9.3000002,
5.3000002, 5.3000002, 9.3000002]
Reference: https://js.tensorflow.org/api/latest/#mirrorPad
Share your thoughts in the comments
Please Login to comment...