본문 바로가기
  • Home

Development of Intelligent Personal Color Diagnosis System by Applying Data Mining Techniques

  • Journal of Knowledge Information Technology and Systems
  • Abbr : JKITS
  • 2017, 12(6), pp.805-815
  • DOI : 10.34163/jkits.2017.12.6.002
  • Publisher : Korea Knowledge Information Technology Society
  • Research Area : Interdisciplinary Studies > Interdisciplinary Research
  • Published : December 31, 2017

jodaseol 1 KIM JUNWOO 1

1동아대학교

Accredited

ABSTRACT

In recent, there is a rising interest on self-presentation based on personal color, a color that matches an individuals’ physical attributes, with an increased emphasis on individuals’ personalities. Nevertheless, it is inconvenient for the individuals to determine their own personal colors accurately, since conventional personal color diagnosis procedures highly depends on various instruments and subjective judgment. In order to address this problem, this paper aims to develop an intelligent personal color diagnosis system that allows the individuals to identify their personal colors by themselves. The proposed system maintains the face images of celebrities with known personal colors, and the personal color of an user is determined by applying the diagnosis model based on the personal color data of celebrities. In this paper, the diagnosis model for the proposed system is constructed by applying the conventional classification techniques such as decision tree, nearest neighbor classifier and naive Bayesian classifier to the face image data of 100 celebrities, and the consequent model is applied to 20 female undergraduates with known personal color identified through visiting professional consultants. The experiment results demonstrate that the proposed system has the promising diagnosis ability so that it can replace the human experts. Moreover, it is expected that the system not only allows the individuals to identify their personal color conveniently, but also enables advanced services based on personal color.

Citation status

* References for papers published after 2023 are currently being built.

This paper was written with support from the National Research Foundation of Korea.