Leveraging Weakly Annotated Data for Fashion Image Retrieval and Label Prediction

TL;DR

This paper introduces a weakly supervised learning approach to develop a visual representation for fashion images, enabling effective clothing classification and image retrieval without manual labels, achieving near state-of-the-art results.

Contribution

The method leverages noisy, automatically crawled data to learn versatile fashion image representations without manual annotation, suitable for classification and retrieval tasks.

Findings

Achieves near state-of-the-art results on DeepFashion tasks

Effective learning from noisy, weakly labeled data

Applicable to multiple fashion image tasks

Abstract

In this paper, we present a method to learn a visual representation adapted for e-commerce products. Based on weakly supervised learning, our model learns from noisy datasets crawled on e-commerce website catalogs and does not require any manual labeling. We show that our representation can be used for downward classification tasks over clothing categories with different levels of granularity. We also demonstrate that the learnt representation is suitable for image retrieval. We achieve nearly state-of-art results on the DeepFashion In-Shop Clothes Retrieval and Categories Attributes Prediction tasks, without using the provided training set.