HKR 手写文本识别数据集

HKR 全称 Handwritten Kazakh and Russian，是关于离线手写文本识别的俄语和哈萨克语表格数据集（约 95% 的俄语和 5% 的哈萨克语）。包含 1,400 多个已填写表格、 63,000 个句子、 715,699 多个字符和 200 位作者。表格由 LATEX 生成，其内容由作者手写。该数据集用 Cyrillic 语言书写，共享 33 个相同的字符。除了这些字符，哈萨克语言的字母还包含 9 个额外的特定字符。

HKR, whose full name is Handwritten Kazakh and Russian, is a tabular dataset for offline handwritten text recognition, comprising approximately 95% Russian content and 5% Kazakh content. It contains over 1,400 completed forms, 63,000 sentences, more than 715,699 characters, and 200 distinct authors. The forms were generated using LaTeX, with their contents handwritten by the authors. This dataset is written in Cyrillic script, and the two languages share 33 identical characters. In addition to these shared characters, the Kazakh language includes 9 additional specific characters.

搜集汇总

背景与挑战

背景概述

HKR手写文本识别数据集是一个专注于俄语和哈萨克语的离线手写文本识别数据集，包含超过1,400份手写表格和63,000个句子，主要用西里尔字母书写。该数据集由社区贡献，涵盖地理关键词和诗歌等手写样本，适用于教育和研究目的。