LOCL: Learning Object-Attribute Composition using Localization

TL;DR

LOCL introduces a modular localization-based approach for object-attribute composition in cluttered scenes, significantly improving zero-shot learning performance and enabling integration with existing methods.

Contribution

It presents a novel modular localization framework that enhances zero-shot object-attribute learning in complex environments, outperforming current state-of-the-art methods.

Findings

12% performance improvement over SOTA in challenging datasets

Modular approach enables integration with existing OA methods

Robust generalization to unseen object-attribute configurations

Abstract

This paper describes LOCL (Learning Object Attribute Composition using Localization) that generalizes composition zero shot learning to objects in cluttered and more realistic settings. The problem of unseen Object Attribute (OA) associations has been well studied in the field, however, the performance of existing methods is limited in challenging scenes. In this context, our key contribution is a modular approach to localizing objects and attributes of interest in a weakly supervised context that generalizes robustly to unseen configurations. Localization coupled with a composition classifier significantly outperforms state of the art (SOTA) methods, with an improvement of about 12% on currently available challenging datasets. Further, the modularity enables the use of localized feature extractor to be used with existing OA compositional learning methods to improve their overall performance.