Research

Paper

AI LLM March 17, 2026

FlowComposer: Composable Flows for Compositional Zero-Shot Learning

Authors

Zhenqi He, Lin Li, Long Chen

Abstract

Compositional zero-shot learning (CZSL) aims to recognize unseen attribute-object compositions by recombining primitives learned from seen pairs. Recent CZSL methods built on vision-language models (VLMs) typically adopt parameter-efficient fine-tuning (PEFT). They apply visual disentanglers for decomposition and manipulate token-level prompts or prefixes to encode compositions. However, such PEFT-based designs suffer from two fundamental limitations: (1) Implicit Composition Construction, where composition is realized only via token concatenation or branch-wise prompt tuning rather than an explicit operation in the embedding space; (2) Remained Feature Entanglement, where imperfect disentanglement leaves attribute, object, and composition features mutually contaminated. Together, these issues limit the generalization ability of current CZSL models. In this paper, we are the first to systematically study flow matching for CZSL and introduce FlowComposer, a model-agnostic framework that learns two primitive flows to transport visual features toward attribute and object text embeddings, and a learnable Composer that explicitly fuses their velocity fields into a composition flow. To exploit the inevitable residual entanglement, we further devise a leakage-guided augmentation scheme that reuses leaked features as auxiliary signals. We thoroughly evaluate FlowComposer on three public CZSL benchmarks by integrating it as a plug-and-play component into various baselines, consistently achieving significant improvements.

Metadata

arXiv ID: 2603.16641
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-17
Fetched: 2026-03-18 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.16641v1</id>\n    <title>FlowComposer: Composable Flows for Compositional Zero-Shot Learning</title>\n    <updated>2026-03-17T15:12:39Z</updated>\n    <link href='https://arxiv.org/abs/2603.16641v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.16641v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Compositional zero-shot learning (CZSL) aims to recognize unseen attribute-object compositions by recombining primitives learned from seen pairs. Recent CZSL methods built on vision-language models (VLMs) typically adopt parameter-efficient fine-tuning (PEFT). They apply visual disentanglers for decomposition and manipulate token-level prompts or prefixes to encode compositions. However, such PEFT-based designs suffer from two fundamental limitations: (1) Implicit Composition Construction, where composition is realized only via token concatenation or branch-wise prompt tuning rather than an explicit operation in the embedding space; (2) Remained Feature Entanglement, where imperfect disentanglement leaves attribute, object, and composition features mutually contaminated. Together, these issues limit the generalization ability of current CZSL models. In this paper, we are the first to systematically study flow matching for CZSL and introduce FlowComposer, a model-agnostic framework that learns two primitive flows to transport visual features toward attribute and object text embeddings, and a learnable Composer that explicitly fuses their velocity fields into a composition flow. To exploit the inevitable residual entanglement, we further devise a leakage-guided augmentation scheme that reuses leaked features as auxiliary signals. We thoroughly evaluate FlowComposer on three public CZSL benchmarks by integrating it as a plug-and-play component into various baselines, consistently achieving significant improvements.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <published>2026-03-17T15:12:39Z</published>\n    <arxiv:comment>Accepted to CVPR2026</arxiv:comment>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Zhenqi He</name>\n    </author>\n    <author>\n      <name>Lin Li</name>\n    </author>\n    <author>\n      <name>Long Chen</name>\n    </author>\n  </entry>"
}