dh-unibe/transkribus-exports-74823-raw-xml

该数据集名为transkribus-exports-74823-raw-xml，是通过Transkribus PageXML数据转换而来。数据集包含148494个训练样本，每个样本包含图像、XML内容、文件名和项目名称等信息。数据以parquet文件形式组织，按项目和分片存储。数据集可能用于图像到文本转换、手写文本识别(HTR)和转录等任务。

This dataset named transkribus-exports-74823-raw-xml was created using pagexml-hf converter from Transkribus PageXML data. It contains 148494 training samples, each including image, XML content, filename and project name. The data is organized as parquet shards by project and split. The dataset is likely used for image-to-text conversion, handwritten text recognition (HTR) and transcription tasks.