MagicLEMP/Hermes-challenge

这是一个用于验证Hermès挑战的小型数据集，包含了与挑战中使用的模型相似的图像。数据集用于图像分类任务，其中的图像分为真实和虚假两类。真实图像是直接由相机捕获的，可能包含一些非视觉内容更改的修改，如水印或新闻标题。虚假图像是完全由计算机生成的，或者是对真实图像进行了内容更改，如使用Photoshop进行编辑或使用图像修复技术。

This is a small dataset for validating the Hermès challenge, containing images similar to those used in the challenge. The dataset is for an image classification task, with images categorized as real or fake. Real images are directly captured by a camera and may have modifications that do not alter the visual content, such as watermarks or news banners. Fake images are either fully computer-generated or have been altered in content, such as editing with Photoshop or using image inpainting techniques.