All-in-One Multilingual Scene Text Recognition with Script-aware Mixture-of-Experts

类型：arxiv
标识：2609.24058
链接：https://arxiv.org/abs/2609.24058
主分类：multimodal
形态：application
TLDR：Multilingual scene text recognition (STR) remains challenging due to the scarcity of training data for most languages and the difficulty of serving diverse scripts within a single model. Existing solutions either deploy one recognizer per language, inflating cost and introducing error accumulation, or rely on massive vision-language models (VLMs) that are expensive and still inaccurate on many scripts. In this work, we pursue an all-in-one multilingual recognizer that is simpler than per-language experts, lighter than VLMs, and more accurate than both. First, we construct TextMuSS-10M, a large
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-23-agent-rag-longcontext-candidates.json

2026-07-12

笔记精读与批判 v2 · Image Tokenizers as Visual Languages in Unified Multimodal Models · 2026-09-12 中午棒（重写覆盖）

2026-09-12

笔记 Cool Papers · cs.CL (papers.cool) · RSS 摘要

2026-09-14

仓库 leon2k2k2k/llm-feedback-alignment

2026-08-11

仓库 Erikalaylafajri15/MOSS-VL

2026-09-22

笔记 flyP 批判性精读 v2 · Vision Encoders in Vision-Language Models: A Survey（Jina / Han Xiao）· 2026-08-25 21:20 CST

2026-08-25

arXiv

2609.24058

TLDR

Multilingual scene text recognition (STR) remains challenging due to the scarcity of training data for most languages and the difficulty of serving diverse scripts within a single model. Existing solutions either deploy one recognizer per language, inflating cost and introducing error accumulation, or rely on massive vision-language models (VLMs) that are expensive and still inaccurate on many scripts. In this work, we pursue an all-in-one multilingual recognizer that is simpler than per-language experts, lighter than VLMs, and more accurate than both. First, we construct TextMuSS-10M, a large

arXiv ↗ 中文全文

All-in-One Multilingual Scene Text Recognition with Script-aware Mixture-of-Experts

相关 · 按语义匹配