convert from one-hot encoding to class label

Question

I have tensor named y, which has values from one-hot encoding over class labels: y = [[ 0. 0. 1. ..., 0. 0. 0.],[ 1. 0. 0. ..., 0. 0. 0.],[ 0. 0. 0. ..., 0. 1. 0.], ...,[ 0. 0. 0. ..., 0. 0. 0.],[ 0. 0. 0. ..., 0. 0. 1.],[ 0. 0. 1. ..., 0. 0. 0.]] so here first row has third element as '1' so it represents class label for that image. Am trying to get all class labels from the given one-hot encoded array, or the given example it should be something like this: y = [2,0,8,...,9,2]

Possible duplicate of [how to convert one hot encodings into integers](http://stackoverflow.com/questions/42497340/how-to-convert-one-hot-encodings-into-integers) — Franck Dernoncourt, Mar 07 '17 at 04:11
You can do that inside tensorflow graph using tf.argmax(y,1) — Harsha Pokkalla, Mar 07 '17 at 14:48

score 0 · Answer 1 · answered Nov 02 '20 at 08:49

0

I think the simplest way is:

import numpy as np
y = np.argmax(y)

answered Nov 02 '20 at 08:49

Владимир Явдошенко

11
1

Without axis = 0 it returns just one number. – MosQuan Sep 29 '22 at 14:05

convert from one-hot encoding to class label

1 Answers1