Learning Attention Propagation for Compositional Zero-Shot Learning

TL;DR

This paper introduces CAPE, a novel attention-based method that models complex dependencies between compositions of visual primitives, significantly improving zero-shot recognition of unseen object-state combinations.

Contribution

The paper proposes CAPE, a new approach that captures rich dependency structures among compositions, enabling better generalization to unseen combinations in compositional zero-shot learning.

Findings

Outperforms previous methods on three benchmarks.

Sets new state-of-the-art in generalized compositional zero-shot learning.

Effectively models complex interactions between primitives.

Abstract

Compositional zero-shot learning aims to recognize unseen compositions of seen visual primitives of object classes and their states. While all primitives (states and objects) are observable during training in some combination, their complex interaction makes this task especially hard. For example, wet changes the visual appearance of a dog very differently from a bicycle. Furthermore, we argue that relationships between compositions go beyond shared states or objects. A cluttered office can contain a busy table; even though these compositions don't share a state or object, the presence of a busy table can guide the presence of a cluttered office. We propose a novel method called Compositional Attention Propagated Embedding (CAPE) as a solution. The key intuition to our method is that a rich dependency structure exists between compositions arising from complex interactions of primitives in addition to other dependencies between compositions. CAPE learns to identify this structure and propagates knowledge between them to learn class embedding for all seen and unseen compositions. In the challenging generalized compositional zero-shot setting, we show that our method outperforms previous baselines to set a new state-of-the-art on three publicly available benchmarks.