Graph-guided Cross-composition Feature Disentanglement for Compositional Zero-shot Learning

TL;DR

This paper introduces a novel graph-guided cross-composition feature disentanglement method for compositional zero-shot learning, leveraging a large pre-trained vision-language model to improve generalization across compositions.

Contribution

It proposes a cross-composition disentanglement approach with a compositional graph and adapters in CLIP, enhancing primitive feature generalization in CZSL.

Findings

Significant performance improvements on three CZSL benchmarks.

Effective disentanglement of primitive features across compositions.

Validation of components through ablation studies.

Abstract

Disentanglement of visual features of primitives (i.e., attributes and objects) has shown exceptional results in Compositional Zero-shot Learning (CZSL). However, due to the feature divergence of an attribute (resp. object) when combined with different objects (resp. attributes), it is challenging to learn disentangled primitive features that are general across different compositions. To this end, we propose the solution of cross-composition feature disentanglement, which takes multiple primitive-sharing compositions as inputs and constrains the disentangled primitive features to be general across these compositions. More specifically, we leverage a compositional graph to define the overall primitive-sharing relationships between compositions, and build a task-specific architecture upon the recently successful large pre-trained vision-language model (VLM) CLIP, with dual cross-composition disentangling adapters (called L-Adapter and V-Adapter) inserted into CLIP's frozen text and image encoders, respectively. Evaluation on three popular CZSL benchmarks shows that our proposed solution significantly improves the performance of CZSL, and its components have been verified by solid ablation studies. Our code and data are available at:https://github.com/zhurunkai/DCDA.