[top]

Worker Qualifications for Image-Aesthetic-Assessment Tasks in Crowdsourcing

by Yudai Kato, Marie Katsurai, Keishi Tajima

Abstract

Image aesthetic assessment has been a trending topic in the research field of multimedia information retrieval. Crowdsourcing can be an efficient approach for collecting manual assessment results to construct an image dataset associated with aesthetic scores. This study explores a strategy for setting worker qualifications to participate in an image-aesthetic-assessment task. Our current experiments based on the AVA dataset indicated that the target subjective task requires highly experienced workers to produce ratings similar to photographers' ones.

Full Text: pdf

Slides: pdf

Poster: pdf

BibTex entry


Work-in-progress paper in HCOMP, #pages: 3, 2022


tajima@i.kyoto-u.ac.jp / Fax: +81(Japan) 75-753-5978 / Office: Research Bldg. #7, room 404