NAF

该数据集包含了来自美国国家档案馆的图像，每一张图像都附有标记的边界框和注释，非常适合用来评估光学字符识别（OCR）的性能。此外，该数据集专为基准测试OCR系统而设计。

This dataset contains images sourced from the National Archives of the United States. Each image is accompanied by labeled bounding boxes and annotations, making it highly suitable for evaluating the performance of Optical Character Recognition (OCR) systems. Furthermore, this dataset is specifically designed for benchmarking OCR systems.