Digital Peter

Digital Peter数据集是由俄罗斯联邦储蓄银行等机构创建，专注于彼得大帝1709至1713年间的历史手稿。该数据集包含9694张图像及对应文本文件，每条记录对应历史文档中的一行。数据集的创建过程涉及从俄罗斯国家古代档案馆获取手稿，并通过计算机视觉注释工具（CVAT）进行手动分割。Digital Peter数据集主要用于训练和评估手写文本识别模型，特别是在处理历史文档中的手写内容时，为研究人员提供了一个重要的基准。

The Digital Peter Dataset was developed by institutions including Sberbank of Russia, focusing on historical manuscripts dating from 1709 to 1713 during the reign of Peter the Great. Comprising 9,694 images and their corresponding text files, each entry in the dataset corresponds to a single line from the historical documents. The dataset creation process involved acquiring manuscripts from the Russian State Archive of Ancient Documents, and manual segmentation was conducted using the Computer Vision Annotation Tool (CVAT). The Digital Peter Dataset is primarily utilized for training and evaluating handwritten text recognition models, and serves as an important benchmark for researchers, particularly when processing handwritten content within historical documents.