From left to right: Songphan Choemprayong (Chula), Shigeo Sugimoto (Tsukuba), Christopher Khoo (Nanyang), Hao-Ren Ke (Taiwan Normal)

Professor Songphan Choemprayong, Ph.D. introduced an academic work under the topic “Applying facial recognition technology to enhance access to a biographical digitized image collection: A case study of princessmcs.org collection” in an Academic conference “Libraries in the Digital Age (LIDA 2016) at Zadar University, Croatia, during 13 – 17 June 2016.

This research work is a cooperation between professors in Department of Library Science in developing digitized image collection research project “Arts Boromrajakumari” in the occasion of Her Royal Highness Princess Maha Chakri Sirindhorn’s 60th Birthday. Interested people may view the project plan at http://www.princessmcs.org

This research project focuses on applying facial recognition technology to enhance access to a biographical digitized image collection with the study of Oral history. This presentation is presented in Panel style in cooperation with leading professors from the School of Library and Information Science in Croatia under the topic “Digital Curation Projects and Research in Asia”.

This research project receives funding from Academics Division of Faculty of Arts, Chulalongkorn University, Department of Library Science Professor Suthilak Ambhanwong’s Fund, and Honorary Consul of Zagreb, Croatia.

บทคัดย่อ/Abstract

One of the major challenges of digitizing images from various amateur sources is a variety of metadata creation practices.  While some sources may apply an exhaustive approach to describe images in their collections, some provide only a small amount of annotations and/or very brief captions.  The later practice seems to be more common among Thai amateur image collectors.  Non-textual information (e.g., persons, objects, and locations) becomes a key element in providing access to biographical digital image collections.  While major open source repository platforms have yet integrated and/or developed ways to process non-textual information, particularly metadata identifying persons in an image.  Thus, we utilized facial recognition feature in Google’s Picasa software to create metadata in identifying persons in a biographical digitized image collection, princessmcs.org collection.  To commemorate the 60th Birthday Anniversary of Her Royal Highness Princess Maha Chakri Sirindhorn, we have collected 2,516 digitized photos representing Her Majesty’s relationship with the Faculty of Arts, Chulalongkorn University, her alma mater, since 1973.  Because most of the photographs in the collection contain less or no description, we implemented oral history approach by arranging 9 rounds of group interviews with Her Royal Highness’ classmates, friends, faculty members and staff members.  There are fourteen participants in total from all 9 rounds.  These participants narrated the stories as well as identified person names in sample images.  Both participants and sample images were selected purposively in order to maximize the efficiency of the identification of persons in the collection.  The sample images were uploaded to Picasa with all other images.  We used the facial recognition feature to group faces that are similar.  The person names given in Picasa were then mapped to Persons element in a Dublin-Core-based repository, operated by Omeka version 2.3.  In total, we identified 583 persons in the collection.  In this paper, we evaluated the performance of facial recognition technology in generating metadata as well as discussed the benefits and drawbacks of this application in a biographical digitized photography collection.

 

Introducing “princessmcs.org” in Croatia

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.