The Recommendation Using Colors and Pictures in Package Design by Integrated Ontology with K-mean Algorithm

Abstract

The packaging design process requires knowledge and methods of thinking, good composition and advice from graphic experts which requires a lot of time and money. Therefore, doing this research to gather knowledge related to packaging design methods from books /textbooks, focus group from Experts and manufacturer. Develop meaningful knowledge base with ontology. Explore and create dataset from product packaging and from customer reviews. Develop the prototype software by dividing into 3 parts which are 1) detecting color and  finding  most color area in the product image using the OpenCV library on Python 2) identifying   the suitable background color  and 3) clustering and recommending  color and  image which decorate on the packaging using integrated ontology with K-mean algorithm. And test the efficiency of clustering and recommendation  of colors and pictures in  package design. The K-mean with  k=2   gives accuracy in clustering  100% and the efficiency of linking the meaning of colors and images has a F-measure equal to 94.0%.