An Automated Make-up Color Selection System

Abstract

The thesis on the research: An Automated Make-up Color Selection System explains the conduct of the research process from problem definition to the evaluation phase. The objective of this research is to explore a novel approach for automating the process of make-up color selection. This is done by using the expertise knowledge of a professional beautician on make-up color selection, for a fuzzy classification and to train an Artificial Neural Network to accomplish the same. To implement this, we collected a set of facial images of ladies with different skin tones. By associating some dress colors with these faces, a beautician decided the suitable make-up colors for the given faces with given dress colors separately. This data set was used to develop the fuzzy sets and to train a neural network. Because of the nature of the entire process which is very subjective, and involves a lot of linguistic ambiguity regarding the skin color, and the complexity involved with defining and classifying colors, we used a fuzzy logic rule set for classifying the skin color. One external factor we consider is the time of the day the make-up is worn. As for the make-up colors, we used 4 sets standardized make-up color palettes for the 4 make-up types foundation, blush, eye-shadows and lipstick, but this could vary according to personnel preference, brands and available colors. With the results, we further elaborate how this approach for make-up selection automation can be enhanced and suggest other schemes where such an approach could be implemented and tested. In the thesis we present a complete overview of the research model, the background knowledge of the make-up selection gathered from the beautician and other resources. The report also describes how the fundamental problems in the proposed method such as data collection, feature extraction, defining of fuzzy sets over the feature sets, the ANN models and the results of the experiments, in detail.