Link to this sectionCIFAR-10 Dataset#

The CIFAR-10 (Canadian Institute For Advanced Research) dataset is a classic image classification benchmark of 60,000 32x32 color images evenly split across 10 classes — airplane, automobile, bird, cat, deer, dog, frog, horse, ship, and truck. It ships with a predefined split of 50,000 training and 10,000 test images (6,000 per class), making it a lightweight, well-balanced starting point for training and benchmarking classification models. For a more fine-grained challenge, see the related CIFAR-100 dataset.

Link to this sectionKey Features#

• CIFAR-10 contains 60,000 color images of 32x32 pixels, evenly divided into 10 classes.
• Each class holds exactly 6,000 images — 5,000 for training and 1,000 for testing — so the dataset is perfectly balanced.
• The 10 classes are airplane, automobile, bird, cat, deer, dog, frog, horse, ship, and truck.
• The dataset ships with a predefined train/test split, so no manual or automatic splitting is required.
• CIFAR-10 is a standard benchmark for image classification and object recognition research.

Link to this sectionDataset Structure#

CIFAR-10 ships with an official, predefined split, so no automatic or manual partitioning is needed:

• Classes: 10 (airplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck)
• Total images: 60,000 (32x32 color)
• Training set: 50,000 images (5,000 per class)
• Test set: 10,000 images (1,000 per class)

Link to this sectionApplications#

CIFAR-10 is widely used to train and evaluate image classification models, from classic Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs) to modern deep architectures. Its small image size and balanced classes make it ideal for rapid experimentation, benchmarking new algorithms, and teaching computer vision fundamentals.

Link to this sectionUsage#

Train a YOLO model on CIFAR-10 for 100 epochs at an image size of 32. For the full list of available arguments, see the Training page and the image classification task guide.

from ultralytics import YOLO

# Load a model
model = YOLO("yolo26n-cls.pt")  # load a pretrained model (recommended for training)

# Train the model
results = model.train(data="cifar10", epochs=100, imgsz=32)

Link to this sectionSample Images and Annotations#

The CIFAR-10 dataset contains color images of various objects, providing a well-structured dataset for image classification tasks. Here are some examples of images from the dataset:

The samples show the diversity of the objects in the CIFAR-10 dataset, underlining the value of a varied dataset for training robust image classification models.

Link to this sectionCitations and Acknowledgments#

If you use the CIFAR-10 dataset in your research or development work, please cite the following paper:

@TECHREPORT{Krizhevsky09learningmultiple,
            author={Alex Krizhevsky},
            title={Learning multiple layers of features from tiny images},
            institution={},
            year={2009}
}

We would like to acknowledge Alex Krizhevsky for creating and maintaining the CIFAR-10 dataset as a valuable resource for the machine learning and computer vision research community. For more information about the CIFAR-10 dataset and its creator, visit the CIFAR-10 dataset website.

Link to this sectionFAQ#

Link to this sectionWhat is the CIFAR-10 dataset used for in machine learning?#

The CIFAR-10 dataset is widely used to train and benchmark image classification and object recognition models. It contains 60,000 32x32 color images evenly spread across 10 classes, and its small size and balanced classes make it a fast, reliable benchmark for algorithms such as Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs).

Link to this sectionHow can I train an Ultralytics YOLO model on the CIFAR-10 dataset?#

To train an Ultralytics YOLO model on CIFAR-10, use the code snippets below. The dataset downloads automatically on first use. For a full list of arguments, see the model Training page.

from ultralytics import YOLO

# Load a model
model = YOLO("yolo26n-cls.pt")  # load a pretrained model (recommended for training)

# Train the model
results = model.train(data="cifar10", epochs=100, imgsz=32)

Link to this sectionHow many classes does the CIFAR-10 dataset have?#

CIFAR-10 has 10 classes — airplane, automobile, bird, cat, deer, dog, frog, horse, ship, and truck — with exactly 6,000 images each, for 60,000 images in total. The classes are mutually exclusive and perfectly balanced, with no overlap between categories.

Link to this sectionHow is the CIFAR-10 dataset split into training and test sets?#

CIFAR-10 ships with a predefined split of 50,000 training images and 10,000 test images, with exactly 5,000 training and 1,000 test images per class. Unlike folder-based classification datasets that Ultralytics splits automatically, CIFAR-10's official partition is used as-is, and the test set serves as the validation split during training by default.

Link to this sectionCan I use Ultralytics Platform for training models on the CIFAR-10 dataset?#

Yes. Ultralytics Platform lets you manage datasets, train image classification models, and deploy them without extensive coding. It is a convenient way to run CIFAR-10 experiments in the cloud, and you can explore more options in our classification datasets overview.