Skip to content

Dataset · uoft-cs

CIFAR-100: 60k Images Across 100 Classes

uoft-cs/cifar100

A 32×32 color image dataset for classification across 100 classes and 20 superclasses, with 50k train and 10k test.

What's inside

CIFAR-100 consists of 60,000 32×32 pixel color images across 100 classes, with 600 images per class. There are 50,000 training images and 10,000 test images; each class has 500 train and 100 test images. Classes are grouped into 20 superclasses, and every image has a fine and coarse label.

What you can do with it

  • Train and compare CNNs or ViTs for image classification
  • Test models on fine- and coarse-classification
  • Use as a benchmark for new architectures
Tasks
Image classification
Languages
English
Size
10K–100K rows
Downloads this month
72.5K
Likes
77

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("uoft-cs/cifar100")