Clothing Classification Using Convolutional Neural Networks

The way people dress is fundamentally tied to social identity, and can offer clues about financial status, social status, tastes, and even culture. An algorithm that can identify clothes can help garment companies understand the profile of potential buyers and focus on targeted niche sales, as well as develop campaigns based on customer tastes. In this context, convolutional neural network models have been shown to be efficient in the task of image classification. This paper explores and analyzes models of convolutional neural networks in the task of classifying parts of clothing through images. The models tested and compared in this paper obtained greater accuracy when compared to non-convolutional models in the literature.