Clothing Classification using the Convolutional Neural Network Inception Model by Paula Esplanad-Mayol
Paper presented during the 2nd International Conference on Information Science and Systems (ICISS 2019) Tokai University, Tokyo, Japan. March 16-19, 2019
Abstract
Convolutional neural network (CNN) is a type of artificial neural network commonly used in analyzing images. Images of clothes are great factors in driving the clothing industry today and trends found in the current fashion styles drive the online clothing economy. This study attempts to create a clothing classification tool with the use of convolutional neural networks. A method for creating an automated clothing classifier was presented, trained from 5600 clothing images in 7 classes using the Inception architecture. The trained model obtained an estimated accuracy of 95%. The model identified different clothing categories with an accuracy of 96.2% in coherence with the estimated accuracy result. It also had a recall of 0.981 and a precision of 1. Future works include improvement in recognition capability by enabling future classifiers to recognize and identify full apparel in a single image.