用自训练CTC对齐修正古籍手写稿的标注错误，提升识别准确率。
CTC Transcription Alignment of the Bullinger Letters: Automatic Improvement of Annotation Quality
- 基于CTC算法动态匹配文字行与转录文本，自动纠正标注缺陷。
- 在PyLaia上使字符错误率降低1.1个百分点，对齐精度显著提升。
- 弱模型反哺强模型，适合古籍文本识别与标注优化任务。
由于手写体差异大、纸张退化及布局标注不足，历史文献的手写文字识别仍具挑战。本文聚焦于16世纪大型信件集Bullinger correspondence中的标注错误，尤其是连字符问题。提出一种基于CTC对齐的自训练方法，利用动态规划与带CTC损失训练的模型输出概率，将完整转录与文字行图像进行匹配。实验表明，该方法在PyLaia上实现1.1个百分点的字符错误率（CER）下降，并显著提升对齐准确率。有趣的是，较弱模型反而产生更精确的对齐结果，支持迭代训练策略。我们发布了100页经人工校正的新子集，以及代码与基准测试。本方法可循环应用，持续优化文本识别流水线中的识别与对齐质量。代码与数据见 https://github.com/andreas-fischer-unifr/nntp。
原文摘要 · Abstract (English)
Handwritten text recognition for historical documents remains challenging due to handwriting variability, degraded sources, and limited layout-aware annotations. In this work, we address annotation errors - particularly hyphenation issues - in the Bullinger correspondence, a large 16th-century letter collection. We introduce a self-training method based on a CTC alignment algorithm that matches full transcriptions to text line images using dynamic programming and model output probabilities trained with the CTC loss. Our approach improves performance (e.g., by 1.1 percentage points CER with PyLaia) and increases alignment accuracy. Interestingly, we find that weaker models yield more accurate alignments, enabling an iterative training strategy. We release a new manually corrected subset of 100 pages from the Bullinger dataset, along with our code and benchmarks. Our approach can be applied iteratively to further improve the CER as well as the alignment quality for text recognition pipelines. Code and data are available via https://github.com/andreas-fischer-unifr/nntp.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。