CODEWITHSUNDEEP
pythonnumpytensorflowartificial-intelligence
helpme
helpme

Reputation: 11

Tensorflow issue with conversion of type numpy.float64 to int

I'm creating a very basic AI with Tensorflow, and am using the code from the official docs/tutorial. Here's my full code:

from __future__ import absolute_import, division, print_function
import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt

fashion_mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()
class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']

train_images = train_images / 255.0
train_labels = train_labels / 255.0

plt.figure(figsize=(10,10))
for i in range(25):
    plt.subplot(5,5,i+1)
    plt.xticks([])
    plt.yticks([])
    plt.grid(False)
    plt.imshow(train_images[i], cmap=plt.cm.binary)
    plt.xlabel(class_names[train_labels[i]])
plt.show()

The issue is on this line: 

plt.xlabel(class_names[train_labels[i]])
TypeError: list indices must be integers or slices, not numpy.float64

No problem, change the numpy.float64 to int using .item()

plt.xlabel(class_names[train_labels[i.item()]])
AttributeError: 'int' object has no attribute 'item'

Was it an int in the first place?

This is running on Python 3.7, with Tensorflow 1.13.1.

Upvotes: 1

Views: 525

Answers (2)

Benales
Benales

Reputation: 9

Normalization (division by 255 in this case) is what is usually necessary to do with features, and not with labels, for labels try to use One hot encoding.

Upvotes: 0

edkeveked
edkeveked

Reputation: 18371

The error is caused by 

train_labels = train_labels / 255.0

train_labels is a ndarray of labels. By dividing it to 255, the resulting ndarray contains float numbers. Therefore a floating number is used as an index for class_names resulting to the first error.

list indices must be integers or slices, not numpy.float64

To convert a numpy array x to int, here is the way to go: x.astype(int). But in this case doing so will create an array with all values to 0.

A fix is to remove the line identified above:

from __future__ import absolute_import, division, print_function
import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt

fashion_mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()
class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']

train_images = train_images / 255.0
# train_labels = train_labels / 255.0

plt.figure(figsize=(10,10))
for i in range(25):
    print(train_labels[i], train_images.shape, train_labels.shape, type(train_labels))
    plt.subplot(5,5,i+1)
    plt.xticks([])
    plt.yticks([])
    plt.grid(False)
    plt.imshow(train_images[i], cmap=plt.cm.binary)
    plt.xlabel(class_names[train_labels[i]])
plt.show()

Upvotes: 1

Related Questions