RVL-CDIP

该数据集包含了320,000张训练图像、40,000张验证图像以及40,000张测试图像，专门用于文档分类任务。评估该任务性能的指标是分类准确率。规模上，数据集提供了总计400,000张图像，涵盖了从训练到测试的全过程。

This dataset consists of 320,000 training images, 40,000 validation images, and 40,000 test images, specifically tailored for document classification tasks. The metric for evaluating performance on this task is classification accuracy. In terms of scale, the dataset provides a total of 400,000 images covering the entire process from training to testing.