Referring Image Segmentation via Register-Aware Feature Selection and Adaptive Adaptation
Mathematical Modeling and Algorithm Application · 30.04.2026 · 8 (3) · 123-135
DOI: 10.54097/perg7629 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematical Modeling and Algorithm Application, online erschienen am 30.04.2026. Verfasst von Xiaozhen Gao. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Referring Image Segmentation (RIS) aims to segment target regions in an image at pixel level based on natural language descriptions. This paper proposes RAFS (Register-Aware Feature Selection and Adaptation), a parameter-efficient framework built upon DINOv3. Three lightweight adapter modules are designed: MLFS for adaptive multi-level feature aggregation via learnable Gaussian weighting, SCSA for efficient cross-modal fusion via depthwise separable convolutions and cross-modal attention, and RAFF for leveraging register tokens’ global context to enhance local features. With only 6.31M additional parameters, RAFS achieves competitive performance on RefCOCO, RefCOCO+, and G-Ref benchmarks.
Zitieren
APA
Gao, X. (2026). Referring Image Segmentation via Register-Aware Feature Selection and Adaptive Adaptation. Mathematical Modeling and Algorithm Application, 8(3), 123-135. https://doi.org/10.54097/perg7629
Vancouver
Gao X. Referring Image Segmentation via Register-Aware Feature Selection and Adaptive Adaptation. Mathematical Modeling and Algorithm Application. 2026;8(3):123-135. doi:10.54097/perg7629
BibTeX
@article{Gao2026,
author = {Gao, Xiaozhen},
title = {Referring Image Segmentation via Register-Aware Feature Selection and Adaptive Adaptation},
journal = {Mathematical Modeling and Algorithm Application},
year = {2026},
volume = {8},
number = {3},
pages = {123--135},
doi = {10.54097/perg7629}
}
EUVIMED