CIFAR-10 and CIFAR-100 are small-scale image datasets consisting of 60,000 low-resolution color images labeled into 10 and 100 classes respectively. Developed by researchers at the University of Toronto, the datasets were designed as standardized benchmarks for image classification research. Images depict everyday objects such as animals, vehicles, and household items, with balanced class distributions.
© 2026 - Copyright AIKosh. All rights reserved.