ARC Ukiyo-e Faces Dataset, based on the ARC Ukiyo-e Portal Database, has been published by the Center for Open Data in the Humanities (CODH)
Based mainly on the Art Research Center's Ukiyo-e Portal Database, a collaborative research group consisting of Yingtao Tian (Google Brain Tokyo) and the members of the ROIS-DS Center for Open Data in the Humanities (CODH)--Tarin Clanuwat, Chikahiko Suzuki, and Asanobu Kitamoto--has published a paper on "Ukiyo-e Analysis and Creativity with Attribute and Geometry Annotation," which will be presented at the 12th International Conference on Computational Creativity.
To take a holistic approach to obtain deep insights into the object and style of Ukiyo-e, the collaborative research group conducted a large-scale, quantitative study on Ukiyo-e, analyzing over 11,000 prints from the ARC Ukiyo-e Portal Database.
As a result, the group has published the ARC Ukiyo-e Faces Dataset, a first-of-its-kind, extensive dataset of faces in Ukiyo-e with coherent labels and annotations created by the automatic extraction of facial regions from Ukiyo-e images using state-of-the-art machine learning models.
With more than 665,000 woodblock prints and curated metadata, the Art Research Center (ARC), Ritsumeikan University, provides one of the largest digital archives of Ukiyo-e prints in the world.
We strive to continue our efforts to build a digital environment for Japanese art and culture that bridges information technology and humanities research and facilitates large-scale, born-digital style research with advanced research results.
Links:
ARC Ukiyo-e Faces Dataset: http://codh.rois.ac.jp/ukiyo-e/face-dataset/