arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新，休息一下吧，周末愉快~~
arXiv 2405.14162cs.CV

利用语义分割掩码与嵌入进行细粒度表格分类

Leveraging Semantic Segmentation Masks with Embeddings for Fine-Grained Form Classification

Brigham Young University（杨百翰大学）

Taylor Archibald, Tony Martinez

阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出结合语义分割与多种深度学习嵌入的表示学习策略，用于细粒度无监督表格分类，并贡献两个新数据集，实验证明该方法能有效提升分类效果。

摘要：

历史文献的高效分类对于家谱学、法律研究和历史学术等领域至关重要，在这些领域中，由于人工分类劳动密集且易出错，对大规模收藏进行手动分类并不现实。为解决这一问题，我们提出了一种表征学习策略，该策略整合了语义分割和深度学习模型（如ResNet、CLIP、文档图像Transformer（DiT）和掩码自编码器（MAE）），以生成无需预定义标签即可捕获文档特征的嵌入。据我们所知，我们是首个在细粒度、无监督表格分类上评估嵌入的方法。为改进这些嵌入，我们提出首先将语义分割作为预处理步骤。我们贡献了两个新数据集——19世纪法国和美国1950年人口普查记录——以展示我们的方法。我们的结果表明，这些不同的嵌入技术在区分相似文档类型方面是有效的，并表明应用语义分割可以极大地改善聚类和分类结果。人口普查数据集可在https://github.com/tahlor/census_forms获取。

英文摘要：

Efficient categorization of historical documents is crucial for fields such as genealogy, legal research, and historical scholarship, where manual classification is impractical for large collections due to its labor-intensive and error-prone nature. To address this, we propose a representational learning strategy that integrates semantic segmentation and deep learning models such as ResNet, CLIP, Document Image Transformer (DiT), and masked auto-encoders (MAE), to generate embeddings that capture document features without predefined labels. To the best of our knowledge, we are the first to evaluate embeddings on fine-grained, unsupervised form classification. To improve these embeddings, we propose to first employ semantic segmentation as a preprocessing step. We contribute two novel datasets$\unicode{x2014}$the French 19th-century and U.S. 1950 Census records$\unicode{x2014}$to demonstrate our approach. Our results show the effectiveness of these various embedding techniques in distinguishing similar document types and indicate that applying semantic segmentation can greatly improve clustering and classification results. The census datasets are available at https://github.com/tahlor/census_forms

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。

↑