Vision Graph Prompting via Semantic Low-Rank Decomposition

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ai, Zixiang, Liu, Zichen, Zhou, Jiahuan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913857111326720
author Ai, Zixiang
Liu, Zichen
Zhou, Jiahuan
author_facet Ai, Zixiang
Liu, Zichen
Zhou, Jiahuan
contents Vision GNN (ViG) demonstrates superior performance by representing images as graph structures, providing a more natural way to capture irregular semantic patterns beyond traditional grid or sequence-based representations. To efficiently adapt ViG to downstream tasks, parameter-efficient fine-tuning techniques like visual prompting become increasingly essential. However, existing prompting methods are primarily designed for Transformer-based models, neglecting the rich topological relationships among nodes and edges in graph-based representations, limiting their capacity to model complex semantics. In this paper, we propose Vision Graph Prompting (VGP), a novel framework tailored for vision graph structures. Our core insight reveals that semantically connected components in the graph exhibit low-rank properties. Building on this observation, we introduce a semantic low-rank prompting method that decomposes low-rank semantic features and integrates them with prompts on vision graph topologies, capturing both global structural patterns and fine-grained semantic dependencies. Extensive experiments demonstrate our method significantly improves ViG's transfer performance on diverse downstream tasks, achieving results comparable to full fine-tuning while maintaining parameter efficiency. Our code is available at https://github.com/zhoujiahuan1991/ICML2025-VGP.
format Preprint
id arxiv_https___arxiv_org_abs_2505_04121
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Vision Graph Prompting via Semantic Low-Rank Decomposition
Ai, Zixiang
Liu, Zichen
Zhou, Jiahuan
Computer Vision and Pattern Recognition
Vision GNN (ViG) demonstrates superior performance by representing images as graph structures, providing a more natural way to capture irregular semantic patterns beyond traditional grid or sequence-based representations. To efficiently adapt ViG to downstream tasks, parameter-efficient fine-tuning techniques like visual prompting become increasingly essential. However, existing prompting methods are primarily designed for Transformer-based models, neglecting the rich topological relationships among nodes and edges in graph-based representations, limiting their capacity to model complex semantics. In this paper, we propose Vision Graph Prompting (VGP), a novel framework tailored for vision graph structures. Our core insight reveals that semantically connected components in the graph exhibit low-rank properties. Building on this observation, we introduce a semantic low-rank prompting method that decomposes low-rank semantic features and integrates them with prompts on vision graph topologies, capturing both global structural patterns and fine-grained semantic dependencies. Extensive experiments demonstrate our method significantly improves ViG's transfer performance on diverse downstream tasks, achieving results comparable to full fine-tuning while maintaining parameter efficiency. Our code is available at https://github.com/zhoujiahuan1991/ICML2025-VGP.
title Vision Graph Prompting via Semantic Low-Rank Decomposition
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.04121