arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2501.12751cs.IRcs.CVcs.LG
收藏

使用大型视觉-语言模型进行专利图形分类

Patent Figure Classification using Large Vision-language Models

TIB – Leibniz Information Centre for Science and Technology(莱布尼茨科技信息中心(TIB));L3S Research Center, Leibniz University(莱布尼茨大学L3S研究中心)

Sushil Awale, Eric Müller-Budack, Ralph Ewerth

AI总结:

本研究探索大型视觉-语言模型在专利图形分类中的应用,提出锦标赛式分类策略及新数据集,在少样本场景下验证了方法的可行性。

摘要:

专利图形分类有助于专利检索系统中的分面搜索,从而实现高效的现有技术检索。现有方法仅针对单一方面的专利图形分类以及概念数量有限的方面进行了探索。近年来,大型视觉-语言模型(LVLMs)在众多计算机视觉下游任务中展现了卓越的性能,然而,它们在专利图形分类方面仍未得到探索。我们的工作探讨了LVLMs在专利图形视觉问答(VQA)和分类中的有效性,重点关注零样本和少样本学习场景。为此,我们引入了新数据集PatFigVQA和PatFigCLS,用于针对专利图形的多个方面(即类型、投影、专利类别和对象)进行微调和评估。为了使用LVLM高效处理大量类别,我们提出了一种新颖的锦标赛式分类策略,该策略利用一系列多项选择题。基于LVLMs和卷积神经网络(CNNs)的多种分类方法在少样本设置下的实验结果和比较显示了所提方法的可行性。

英文摘要:

Patent figure classification facilitates faceted search in patent retrieval systems, enabling efficient prior art search. Existing approaches have explored patent figure classification for only a single aspect and for aspects with a limited number of concepts. In recent years, large vision-language models (LVLMs) have shown tremendous performance across numerous computer vision downstream tasks, however, they remain unexplored for patent figure classification. Our work explores the efficacy of LVLMs in patent figure visual question answering (VQA) and classification, focusing on zero-shot and few-shot learning scenarios. For this purpose, we introduce new datasets, PatFigVQA and PatFigCLS, for fine-tuning and evaluation regarding multiple aspects of patent figures~(i.e., type, projection, patent class, and objects). For a computational-effective handling of a large number of classes using LVLM, we propose a novel tournament-style classification strategy that leverages a series of multiple-choice questions. Experimental results and comparisons of multiple classification approaches based on LVLMs and Convolutional Neural Networks (CNNs) in few-shot settings show the feasibility of the proposed approaches.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑