Limerencii/russian-handwriting-ocr

这是一个俄文手写文本识别数据集，包含手写文本的图像和对应的文本内容。数据集旨在用于对vision-language模型（如Qwen3 VL）进行微调，以识别俄文手写文本。数据集共有13050个样本，分为训练集、验证集和测试集。图像类型包括扫描件、暗光照片和亮光照片。每个样本包含一个图像和对话格式的信息，其中包含用户请求和助手回复的文本。

This is a Russian handwritten text recognition dataset containing images of handwritten texts and their corresponding transcriptions. The dataset is intended for fine-tuning vision-language models (such as Qwen3 VL) to recognize Russian handwritten text. It includes 13,050 samples divided into training, validation, and test sets. Image types include scanned documents, darklit photos, and lightlit photos. Each sample consists of an image and conversational formatted information, including a user request and an assistants response text.