发表机构
TU Wien; Austrian Archaeological Institute; Austrian Academy of Sciences（维也纳技术大学; 奥地利考古研究所; 奥地利科学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究推出CENTURIA陶器元数据数据集，发现OCR模型处理手写陶器文档存在显著领域差距，经少量样本LoRA微调后可大幅提升转录与字段提取性能，为考古数字化提供可行方案。
AI 中文摘要
陶器是重建古代社会时间维度与经济维度的主要来源。考古学家常通过技术绘图与手写元数据记录陶器发现，这些元数据对断代、溯源归属及跨遗址比对至关重要，但目前无法用于计算分析，需人工转录每条记录。本研究探究最先进的文档分析模型能否解决该任务，并推出CENTURIA数据集，包含来自古罗马卡农图姆遗址的507条陶器记录，提供转录文本、边界框及7个元数据类别的结构化字段标签。对5种OCR模型的基准测试显示存在显著领域差距：零样本转录误差达15-32% SpACER-M，远超印刷档案文献的误差率，领域特定字段的恢复率不足3%。仅用57个样本进行LoRA微调（反映实际档案标注预算）即可缩小该差距，将转录误差降至1.5%以下，整体字段级准确率恢复至87%以上。研究表明，少量经专家验证的微调数据集足以将手写陶器文档转换为结构化、可搜索的元数据，可直接用于考古数据库。
英文摘要
Pottery is a primary source for reconstructing the chronological and economic dimensions of past societies. Archaeologists often document ceramic finds through technical drawings and handwritten metadata. This metadata is critical for dating, provenance attribution, and cross-site comparison, but remains inaccessible to computational analysis, requiring manual transcription of every record. We investigate whether state-of-the-art document analysis models can address this task, and introduce CENTURIA, a dataset of 507 pottery records from the Roman site of Carnuntum, providing transcriptions, bounding boxes, and structured field-level labels across seven metadata categories. Benchmarking five OCR models reveals a substantial domain gap: zero-shot transcription error reaches 15-32% SpACER-M, far exceeding rates on printed archival documents, with domain-specific fields recovered in fewer than 3% of cases. LoRA fine-tuning on just 57 samples, reflecting a realistic archival annotation budget, closes this gap, reducing transcription error to below 1.5% and recovering overall field-level accuracy above 87%. Our results show that a small expert-validated fine-tuning set suffices to convert handwritten pottery documentation into structured, searchable metadata ready for archaeological databases.
