WITHDRAWN: Fashion compatibility using convolutional neural networks

Abstract

Abstract The exponential growth in online shopping have increased the sales in apparel industry. The underlying secret is the online fashion recommendation system, which gives the customer various suggestions on the outfit selection during online shopping. These days fashionable outfit is the first wish and being able to choose from a wide variety of apparel and accessory combination is trending. The customers attain satisfaction by browsing through a variety of recommended outfits along with their associated accessories during online shopping. In this paper, we develop a fashion compatibility system that suggests the user with the outfit and its associated accessories for given input text like a fashion designer suggestion. We address the problem of generating fashion compatible outfits and accessories using Convolutional Neural Networks (CNN). The fashion compatibility system investigates for the simplest technique to recommend fashion compatible products that helps the retailers to understand the sentiment analysis of the customers in order to extend their digital marketing and customer satisfaction. We have compared different feature extraction techniques like bag of words, TDF-IDF, word2vec model. The model we have used to train the dataset is VGG-16.