flwrlabs/pacs

--- dataset_info: features: - name: image dtype: image - name: domain dtype: string - name: label dtype: class_label: names: '0': dog '1': elephant '2': giraffe '3': guitar '4': horse '5': house '6': person splits: - name: train num_bytes: 252282893.372 num_examples: 9991 download_size: 191395900 dataset_size: 252282893.372 configs: - config_name: default data_files: - split: train path: data/train-* license: unknown task_categories: - image-classification size_categories: - 1K<n<10K --- # 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](https://flower.ai/docs/datasets/) (flwr-datasets) for the dataset download and partitioning and [Flower](https://flower.ai/docs/framework/) (flwr) for conducting FL experiments. To partition the dataset, do the following. 1. Install the package. ```bash pip install flwr-datasets[vision] ``` 2. Use the HF Dataset under the hood in Flower Datasets. ```python 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](https://flower.ai/).

搜集汇总

背景与挑战

背景概述

PACS数据集是一个用于域泛化（Domain Generalization）的图像分类数据集，包含四个领域（照片、艺术绘画、卡通、素描），每个领域有7个类别，共9991个样本。该数据集以Parquet格式提供，并特别适配了联邦学习框架Flower，便于进行分布式训练研究。