Ukiyo-e Face Datasets

Introduce methodologies of machine learning and data science to Ukiyo-e research, and construct a new digital research infrastructure on Japanese culture.

ARC Ukiyo-e Faces Dataset

ARC Ukiyo-e Faces Dataset is the dataset of faces in Ukiyo-e created by the automatic extraction of facial regions from Ukiyo-e images using machine learning. The dataset is created by the collaborative research group and is based on "Ritsumeikan ARC Ukiyo-e Database" (Art Research Center (ARC), Ritsumeikan University) via Informatics Research Data Repository (IDR) of National Institute of Informatics. The following GitHub repository provides both the dataset of faces and scripts for downloading images and analyzing them.

GitHub: rois-codh/arc-ukiyoe-faces: ARC Ukiyo-e Faces Dataset

Figure: Ten creators with the largest number of Ukiyo-e images in the dataset.

As of June 2021, the latest version (v1.0) of the dataset contains 16653 Ukiyo-e face datas extracted from 9203 Ukiyo-e images.

Please consider citing the following papers when publishing your research results using ARC Ukiyo-e Faces Dataset.

Yingtao Tian, Tarin Clanuwat, Chikahiko Suzuki, Asanobu Kitamoto, "Ukiyo-e Analysis and Creativity with Attribute and Geometry Annotation", arXiv:2106.02267, 2021.

Please refer to License / Citation and Guidelines before using the dataset.

Summary of the Dataset

ARC Ukiyo-e Faces Dataset consists of two types of data as follows.

Metadata ARC assigned metadata to the original images, including bibliographic information such as title, player, publisher, creator and time.
Annotation data The collaborative research group compiled the result of automatic extraction of facial regions and facial components (e.g. eye, mouth and nose) using machine learning.

List of Dataset Creators and Contributors

Creators

Collaborative research group
  • Yingtao Tian (Google Brain Tokyo)
  • Tarin Clanuwat (ROIS-DS Center for Open Data in the Humanities / NII)
  • Chikahiko Suzuki (ROIS-DS Center for Open Data in the Humanities / NII)
  • Asanobu Kitamoto (ROIS-DS Center for Open Data in the Humanities / NII)

Contributors

Release of metadata and images
Distribution of metadata

License / Citation

Creative Commons License
"ARC Ukiyo-e Faces Dataset" (Created by Yingtao Tian and ROIS-DS CODH; Collected from ARC) is licensed under Creative Commons Attribution 4.0 International License (CC BY).

When you publish your creative work (e.g. paper), the following credit is required under the CC BY license.

When you use the annotation data

Please cite the following.

"ARC Ukiyo-e Faces Dataset" (Created by Yingtao Tian, ROIS-DS CODH; Collected from ARC), https://doi.org/10.20676/00000394

When you use the metadata or images

Please cite the following in addition to above.

Art Research Center, Ritsumeikan University (2020): ARC Ukiyo-e database. Informatics Research Data Repository, National Institute of informatics. (dataset). https://doi.org/10.32130/rdata.2.1

Here the metadata is distributed within the ARC Ukiyo-e Faces Dataset with a permission from Art Research Center, Ritsumeikan University.

When you refer to research on the dataset

Please cite the following in addition to above.

Yingtao Tian, Tarin Clanuwat, Chikahiko Suzuki, Asanobu Kitamoto, "Ukiyo-e Analysis and Creativity with Attribute and Geometry Annotation", arXiv:2106.02267, 2021.

Mahalo Button

What is Mahalo Button?

Usage Guidelines

ARC Ukiyo-e Faces Dataset (this dataset) provides the collection of facial expression images cropped from Ukiyo-e publicly available from Art Research Center, Ritsumeikan University. Please follow the guidelines when you use this dataset.

  1. This dataset may contain entities that are respected in religion, ideology, or for other reasons. Please respect diverse values and avoid degrading the respected subject.
  2. Please respect the original works, creators and providers of this dataset. We believe that a proper credit to the contribution of creators and providers is essential to promote open data movement.
  3. Public Domain Usage Guidelines - Europeana Collections is also useful as the guidelines for using works created in the past.

Guidelines are based on goodwill. They are not legal contract.

  1. KaoKore Dataset

Latest News

2021-06-07

ARC Ukiyo-e Faces Dataset was released.