The CIFAR-10 machine-learning dataset contains 10 labeled image classes.
CIFAR-10 consists of 60,000 color images, each measuring 32 by 32 pixels. The images are divided into 10 categories, including airplane, automobile, bird, cat, deer, dog, frog, horse, ship, and truck. It was collected as a smaller, more accessible benchmark for object recognition research.
The dataset contains 50,000 training images and 10,000 test images. Because the images are tiny and the classes are familiar, CIFAR-10 became a standard way to compare convolutional networks and other machine-learning methods without requiring the enormous computing resources associated with larger datasets. It is often confused with CIFAR-100, which has 100 classes and only 600 images per class. CIFAR-10 is also distinct from MNIST, whose grayscale images represent handwritten digits rather than everyday color objects.