Learning Attributes Equals Multi-Source Domain Generalization

TL;DR

This paper introduces a novel approach to attribute detection by applying multi-source domain generalization techniques, aiming to improve cross-category robustness and accuracy in attribute detectors across unseen categories.

Contribution

It proposes a new perspective that treats attribute detection as a multi-source domain generalization problem, addressing the challenge of cross-category attribute detection.

Findings

Effective attribute detectors that generalize well across unseen categories.

Significant improvements over existing methods on four challenging datasets.

Versatile approach applicable to multiple computer vision tasks.

Abstract

Attributes possess appealing properties and benefit many computer vision problems, such as object recognition, learning with humans in the loop, and image retrieval. Whereas the existing work mainly pursues utilizing attributes for various computer vision problems, we contend that the most basic problem---how to accurately and robustly detect attributes from images---has been left under explored. Especially, the existing work rarely explicitly tackles the need that attribute detectors should generalize well across different categories, including those previously unseen. Noting that this is analogous to the objective of multi-source domain generalization, if we treat each category as a domain, we provide a novel perspective to attribute detection and propose to gear the techniques in multi-source domain generalization for the purpose of learning cross-category generalizable attribute detectors. We validate our understanding and approach with extensive experiments on four challenging datasets and three different problems.