saberzl/SID_Set

SID_Set是一个社交媒体图像检测数据集，包含30万张由AI生成或篡改的图像以及真实的图像，并具备详尽的注释。数据集的优势在于：拥有大量的图像、涵盖各种类别的全合成和篡改图像、以及高真实性的图像，这些图像通过视觉检查几乎无法与真实图像区分。数据集包含图像ID、图像、掩码、宽度和高度等信息，其中掩码是用于突出篡改图像中的篡改区域的二进制掩码。图像标签分为三类：真实图像、全合成图像和篡改图像。

The SID_Set is a social media image detection dataset that consists of 300K AI-generated/tampered and authentic images with comprehensive annotations. The dataset offers three key advantages: extensive volume, broad diversity with fully synthetic and tampered images across various classes, and elevated realism with images that are predominantly indistinguishable from genuine ones through mere visual inspection. The dataset includes image ID, image, mask, width, and height information, where the mask is a binary mask highlighting manipulated regions in tampered images. The image labels are categorized into three types: real images, full synthetic images, and tampered images.