AI 大模型

视觉大模型 / VLM

视觉语言模型、视觉推理、视觉问答、图文理解和视觉 grounding。

2026-01-12 至 2026-01-12 共收录 2 篇信号源：cs.CV, cs.AI, cs.LG

1. 文档图表理解 2 篇

2601.05688 2026-01-12 cs.CV 77%

SketchVL: Policy Optimization via Fine-Grained Credit Assignment for Chart Understanding and More

SketchVL：通过细粒度信用分配进行政策优化以实现图表理解和更多

Muye Huang, Lingling Zhang, Yifei Li, Yaqiang Wu, Jun Liu

机构 * School of Computer Science and Technology, Xi’an Jiaotong University, China（计算机科学与技术学院，西安交通大学）； MOE KLINNS Lab, Xi’an Jiaotong University, China（教育部KLINNS实验室，西安交通大学）； Zhongguancun Academy, Beijing, China（中关村学院，北京）

专题命中文档图表理解：visual reasoning(abstract)；multimodal large language model(abstract)；MLLM(abstract)；分类 cs.CV

AI总结 SketchVL通过细粒度信用分配优化，提升多模态大语言模型在图表理解和多领域推理中的性能。

2505.10055 2026-01-12 cs.CV cs.AI 62%

PsOCR: Benchmarking Large Multimodal Models for Optical Character Recognition in Low-resource Pashto Language

PsOCR：用于低资源帕什托语言光学字符识别的大型多模态模型基准测试

Ijazul Haq, Yingjie Zhang, Irfan Ali Khan

机构 * organization= Shien-Ming Wu School of Intelligent Manufacturing, South China University of Technology , city= Guangzhou , postcode= 511442 , country= China ； organization= Artificial Intelligence Department, Guangdong CAS Angels Biotechnology Co. Ltd. , city= Foshan , country= China ； organization= Department of Software Engineering, Faculty of Electrical \& Computer Engineering, University of Engineering \& Technology , city= Peshawar , postcode= 25000 , country= Pakistan

专题命中文档图表理解：InternVL(abstract)；分类 cs.CV、cs.AI

AI总结 PsOCR通过合成数据评估大型多模态模型在低资源帕什托语言OCR中的性能，发现Gemini表现最佳，Qwen-7B在开源模型中表现突出。