Dataset Card for PACS

PACS is an image dataset for domain generalization. It consists of four domains, namely Photo (1,670 images), Art Painting (2,048 images), Cartoon (2,344 images), and Sketch (3,929 images). Each domain contains seven categories (labels): Dog, Elephant, Giraffe, Guitar, Horse, and Person. The total number of sample is 9991.

Dataset Details

PACS DG dataset is created by intersecting the classes found in Caltech256 (Photo), Sketchy (Photo, Sketch), TU-Berlin (Sketch) and Google Images(Art painting, Cartoon, Photo).

Dataset Sources

• Website: https://sketchx.eecs.qmul.ac.uk/downloads/
• Paper: https://arxiv.org/pdf/1710.03077
• Papers with code: https://paperswithcode.com/dataset/pacs

Use in FL

In order to prepare the dataset for the FL settings, we recommend using Flower Dataset (flwr-datasets) for the dataset download and partitioning and Flower (flwr) for conducting FL experiments.

To partition the dataset, do the following.

1. Install the package.

pip install flwr-datasets[vision]

2. Use the HF Dataset under the hood in Flower Datasets.

from flwr_datasets import FederatedDataset
from flwr_datasets.partitioner import IidPartitioner

fds = FederatedDataset(
    dataset="flwrlabs/pacs",
    partitioners={"train": IidPartitioner(num_partitions=10)}
)
partition = fds.load_partition(partition_id=0)

Dataset Structure

Data Instances

The first instance of the train split is presented below:

{
 'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=227x227>,
 'domain': 'art_painting',
 'label': 0
}

Data Split

DatasetDict({
    train: Dataset({
        features: ['image', 'domain', 'label'],
        num_rows: 9991
    })
})

Citation

When working with the PACS dataset, please cite the original paper. If you're using this dataset with Flower Datasets and Flower, cite Flower.

BibTeX:

Original paper:

@misc{li2017deeperbroaderartierdomain,
      title={Deeper, Broader and Artier Domain Generalization}, 
      author={Da Li and Yongxin Yang and Yi-Zhe Song and Timothy M. Hospedales},
      year={2017},
      eprint={1710.03077},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/1710.03077}, 
}

Flower:

@article{DBLP:journals/corr/abs-2007-14390,
  author       = {Daniel J. Beutel and
                  Taner Topal and
                  Akhil Mathur and
                  Xinchi Qiu and
                  Titouan Parcollet and
                  Nicholas D. Lane},
  title        = {Flower: {A} Friendly Federated Learning Research Framework},
  journal      = {CoRR},
  volume       = {abs/2007.14390},
  year         = {2020},
  url          = {https://arxiv.org/abs/2007.14390},
  eprinttype    = {arXiv},
  eprint       = {2007.14390},
  timestamp    = {Mon, 03 Aug 2020 14:32:13 +0200},
  biburl       = {https://dblp.org/rec/journals/corr/abs-2007-14390.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}

Dataset Card Contact

If you have any questions about the dataset preprocessing and preparation, please contact Flower Labs.