CIFAR-10

CIFAR-10 computer-vision training dataset

The CIFAR-10 dataset contains 50,000 training and 10,000 test images of 10 object classes ("airplane", "automobile", "bird", "cat", "deer", "dog", "frog", "horse", "ship", "truck"). Each image is an RGB image of size 32x32.

Examples

Open in Wolfram Cloud

Download Example Notebook

Basic Examples

Retrieve the resource:

In[1]:=

Out[1]=

View the data:

In[2]:=

Out[2]=

Analysis

Obtain a random sample of training data:

In[3]:=

Out[3]=

Train a classifier:

In[4]:=

Out[4]=

Obtain general information about the classifier:

In[5]:=

Out[5]=

Obtain a random sample of test data:

In[6]:=

Out[6]=

Generate a ClassifierMeasurementsObject of the classifier with the test set:

In[7]:=

Out[7]=

Visualize the accuracy of the classifier:

In[8]:=

Out[8]=

Bibliographic Citation

Wolfram Research, "CIFAR-10" from the Wolfram Data Repository (2018) https://doi.org/10.24097/wolfram.83212.data

Data Resource History

Date Created: 24 May 2016
Updated: 23 February 2018

Source Metadata

Title: Learning Multiple Layers of Features from Tiny Images
Creator: Alex Krizhevsky
Publisher: Technical report, University of Toronto
Date: 2009
Source: https://www.cs.toronto.edu/~kriz/cifar.html

Publisher Information

Publisher of Record: Wolfram Research