Dynamic Visual-semantic Alignment for Zero-shot Learning with Ambiguous Labels

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Jiangnan, Huang, Linqing, Yan, Xiaowen, Gan, Min, Lu, Wenpeng, Fan, Jinfu
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913045950758912
author Li, Jiangnan
Huang, Linqing
Yan, Xiaowen
Gan, Min
Lu, Wenpeng
Fan, Jinfu
author_facet Li, Jiangnan
Huang, Linqing
Yan, Xiaowen
Gan, Min
Lu, Wenpeng
Fan, Jinfu
contents Zero-shot learning (ZSL) aims to recognize unseen classes without visual instances. However, existing methods usually assume clean labels, overlooking real-world label noise and ambiguity, which degrades performance. To bridge this gap, we propose the Dynamic Visual-semantic Alignment (DVSA), a robust ZSL framework for learning from ambiguous labels. DVSA uses a bidirectional visual-semantic alignment module with attention to mutually calibrate visual features and attribute prototypes, and a contrastive optimization grounded in Mutual Information (MI) at the attribute level to strengthen discriminative, semantically consistent attributes. In addition, a dynamic label disambiguation mechanism iteratively corrects noisy supervision while preserving semantic consistency, narrowing the instance-label gap, and improving generalization. Extensive experiments on standard benchmarks verify that DVSA achieves stronger performance under ambiguous supervision.
format Preprint
id arxiv_https___arxiv_org_abs_2604_17710
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Dynamic Visual-semantic Alignment for Zero-shot Learning with Ambiguous Labels
Li, Jiangnan
Huang, Linqing
Yan, Xiaowen
Gan, Min
Lu, Wenpeng
Fan, Jinfu
Computer Vision and Pattern Recognition
Zero-shot learning (ZSL) aims to recognize unseen classes without visual instances. However, existing methods usually assume clean labels, overlooking real-world label noise and ambiguity, which degrades performance. To bridge this gap, we propose the Dynamic Visual-semantic Alignment (DVSA), a robust ZSL framework for learning from ambiguous labels. DVSA uses a bidirectional visual-semantic alignment module with attention to mutually calibrate visual features and attribute prototypes, and a contrastive optimization grounded in Mutual Information (MI) at the attribute level to strengthen discriminative, semantically consistent attributes. In addition, a dynamic label disambiguation mechanism iteratively corrects noisy supervision while preserving semantic consistency, narrowing the instance-label gap, and improving generalization. Extensive experiments on standard benchmarks verify that DVSA achieves stronger performance under ambiguous supervision.
title Dynamic Visual-semantic Alignment for Zero-shot Learning with Ambiguous Labels
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.17710