Attribute alignment networks for generalized zero-shot learning

Nannan Lu

doi:10.1016/j.patrec.2025.09.010

Attribute alignment networks for generalized zero-shot learning

2025-10-06

0

PRE

AI

DOI：10.1016/j.patrec.2025.09.010

Original

Original request for help

Save

Abstract

En 中文

• A novel attribute alignment pipeline is designed for exploiting multiple attribute information comprehensively in order to gen- erate visual features corresponding to the attributes. • An attribute enhancement module is created to gen- erate enhanced attribute representations to mitigate the impact of intra-class variations. • Extensive experiments on four benchmarks show that our design can significantly improve the results in both ZSL and GZSL tasks.

Attribute alignment networks for generalized zero-shot learning

Abstract

Journal

Organization