Clothing Image Recognition Based on Multiple Features Using Deep Neural Networks

Abstract

For many online customers, clothing image recognition is used mainly in computer vision for fashion applications. Recognition of the clothing image and identification of their style and presentation make the problem difficult for the fashion item. Image recognition technologies allow consumers to scan a picture taken from a fashion magazine or print ad and automatically land on the product page where they can purchase the exact item. Retailers can propose similar looking items in different price ranges with visual search solutions running in the app or website, so the customer can buy a look-alike product at a lower price. The main focus is on classifying the MNIST fashion dataset using multilayer perceptron, convolutional neural network and extreme learning machine. Extracting the F-mnist dataset using these deep neural networks that are most common in computer vision for the implementation of image recognition and even more effective for cloth prediction evaluation. Fashion-MNIS T enhances cultural diversity by attracting more young women students, collectors, artists and designers. Selecting the right deep learning technique for extracting features and choosing the best model of classification remains a major challenge in achieving good quality accuracy. The experimental results, however, show that this Fashion-MNIST dataset has impressive results.