Oxford-IIIT Pet

Overview

The Oxford-IIIT Pet dataset is a 37-category pet image dataset (12 cat breeds, 25 dog breeds) with roughly 200 images per class. Images vary widely in scale, pose, and lighting, making it a standard fine-grained classification and linear-probe transfer benchmark.

Train (trainval): 3,680 images
Test: 3,669 images

This builder only consumes the per-image breed label; the upstream annotation archive also ships trimap segmentations and bounding boxes, which are not exposed here.

../../_images/oxford_iiit_pet_teaser.png

Data Structure

When accessing an example using ds[i], you will receive a dictionary with the following keys:

Key	Type	Description
`image`	`PIL.Image.Image`	RGB image (variable dimensions)
`label`	int	Breed label (0-36), equal to the official class id minus 1

Usage Example

from stable_datasets.images.oxford_iiit_pet import OxfordIIITPet

ds = OxfordIIITPet(split="train")

sample = ds[0]
print(sample.keys())  # {"image", "label"}

class_names = ds.info.features["label"].names
print(class_names[sample["label"]])

# Optional: make it PyTorch-friendly
ds_torch = ds.with_format("torch")

References

Official website: https://www.robots.ox.ac.uk/~vgg/data/pets/
Paper: Cats and Dogs (CVPR 2012)
License: CC BY-SA 4.0 (copyright of the images remains with their original owners)

Citation

@inproceedings{parkhi12a,
  title={Cats and Dogs},
  author={Parkhi, O. M. and Vedaldi, A. and Zisserman, A. and Jawahar, C. V.},
  booktitle={IEEE Conference on Computer Vision and Pattern Recognition},
  year={2012}
}