Voxel51/synthetic_us_passports_easy

这是一个包含9750张合成美国护照图像的数据集，旨在用于文档理解任务的视觉语言模型的训练和评估，以及OCR和信息提取系统的测试。数据集中的每张图像都包含完整的生物识别和文档信息字段。该数据集是合成的，确保没有使用任何真实个人信息，并遵循Apache-2.0许可证。

This dataset consists of 9750 synthetic US passport images designed for training and evaluating Vision Language Models on document understanding tasks, as well as testing OCR and information extraction systems. Each image in the dataset contains complete biographical and document information fields. The dataset is synthetic, ensuring no real personal information is used, and is licensed under Apache-2.0.