Searcher for Clothes on the Web Using Convolutional Neural Networks and Dissimilarity Rules for Color Classification Using Euclidean Distance to Color Centers in the HSL Color Space

Searching and sorting objects, especially items of interest, on the web is an essential task in the digital age. Systems that facilitate this task are of great use to consumers and e-commerce companies. The use of convolutional neural networks (CNN) is a very useful tool when classifying many objects in different semantic fields according to their characteristics. If we add a module for obtaining the color of the base object to this classification and combine both results in the network, we will obtain as a result a system for searching for objects that is not only based on their characteristics but also on their exact color. In this study, we propose a literature-based methodology for obtaining clothing on the web. This system is based on the creation of a color base with different centers in the HSL (Hue, Saturation, Luminosity) space to obtain the color of an image based on dissimilarity rules. The base image previously passed through the convolutional neural network, plus the result of obtaining the color through the Euclidean distance will generate a character string, which will be placed in a search engine and will bring as a result similar garments to the base image. This work provides new perspectives and useful techniques for obtaining colors, which can significantly improve the accuracy and efficiency of object search on the web.