Linguistically Informed Multimodal Fusion for Vietnamese Scene-Text Image Captioning: Dataset, Graph Framework, and Phonological Attention

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Nguyen, Nhi Ngoc-Yen, Nguyen, Anh-Duc, Nguyen, Nghia Hieu, Van Nguyen, Kiet, Nguyen, Ngan Luu-Thuy
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917450631610368
author Nguyen, Nhi Ngoc-Yen
Nguyen, Anh-Duc
Nguyen, Nghia Hieu
Van Nguyen, Kiet
Nguyen, Ngan Luu-Thuy
author_facet Nguyen, Nhi Ngoc-Yen
Nguyen, Anh-Duc
Nguyen, Nghia Hieu
Van Nguyen, Kiet
Nguyen, Ngan Luu-Thuy
contents Scene-text image captioning requires fusing three information streams -- visual features, OCR-detected text, and linguistic knowledge -- to generate descriptions that faithfully integrate text visible in images. Existing fusion approaches treat text as language-agnostic, which fails for Vietnamese: a tonal language where diacritics alter word meaning, OCR errors are pervasive, and word boundaries are ambiguous. We argue that Vietnamese scene-text captioning demands \textit{linguistically informed multimodal fusion}, where language-specific structural knowledge is explicitly incorporated into the fusion mechanism. Motivated from these insights, we propose \textbf{HSTFG} (Heterogeneous Scene-Text Fusion Graph), a general-purpose graph fusion framework with learned spatial attention bias, and show through topology analysis that cross-modal graph edges are harmful for scene-text fusion. Building on this finding, we design \textbf{PhonoSTFG} (Phonological Scene-Text Fusion Graph) which specializes graph-level fusion for Vietnamese linguistic reasoning. To support evaluation, we introduce \textbf{ViTextCaps}, the first large-scale Vietnamese scene-text captioning dataset (\textbf{15{,}729} images with \textbf{74{,}970} captions), with comprehensive linguistic analysis showing that 52.8\% of the vocabulary is at risk of diacritic collision.
format Preprint
id arxiv_https___arxiv_org_abs_2604_27712
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Linguistically Informed Multimodal Fusion for Vietnamese Scene-Text Image Captioning: Dataset, Graph Framework, and Phonological Attention
Nguyen, Nhi Ngoc-Yen
Nguyen, Anh-Duc
Nguyen, Nghia Hieu
Van Nguyen, Kiet
Nguyen, Ngan Luu-Thuy
Computer Vision and Pattern Recognition
Computation and Language
Scene-text image captioning requires fusing three information streams -- visual features, OCR-detected text, and linguistic knowledge -- to generate descriptions that faithfully integrate text visible in images. Existing fusion approaches treat text as language-agnostic, which fails for Vietnamese: a tonal language where diacritics alter word meaning, OCR errors are pervasive, and word boundaries are ambiguous. We argue that Vietnamese scene-text captioning demands \textit{linguistically informed multimodal fusion}, where language-specific structural knowledge is explicitly incorporated into the fusion mechanism. Motivated from these insights, we propose \textbf{HSTFG} (Heterogeneous Scene-Text Fusion Graph), a general-purpose graph fusion framework with learned spatial attention bias, and show through topology analysis that cross-modal graph edges are harmful for scene-text fusion. Building on this finding, we design \textbf{PhonoSTFG} (Phonological Scene-Text Fusion Graph) which specializes graph-level fusion for Vietnamese linguistic reasoning. To support evaluation, we introduce \textbf{ViTextCaps}, the first large-scale Vietnamese scene-text captioning dataset (\textbf{15{,}729} images with \textbf{74{,}970} captions), with comprehensive linguistic analysis showing that 52.8\% of the vocabulary is at risk of diacritic collision.
title Linguistically Informed Multimodal Fusion for Vietnamese Scene-Text Image Captioning: Dataset, Graph Framework, and Phonological Attention
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2604.27712