Tocmac/ReinAD

ReinAD数据集是一个全面的对比数据集，旨在实现现实世界的工业异常检测。该数据集包含训练集和测试集，所有正常和异常图像都存储为hdf5格式。在掩码注释中，值为0的像素代表正常区域，值为1的像素代表异常区域。数据集的文件结构是统一的，包含不同类别的h5文件。每个hdf5文件内部包含图像和对应的掩码，其中图像分为异常和正常两种类型，掩码为像素级别的异常区域注释。

The ReinAD dataset is a comprehensive contrastive dataset designed for real-world industrial anomaly detection. It consists of a training set and a test set, with all normal and anomaly images stored in hdf5 format. In the mask annotations, pixels with a value of 0 represent normal regions, and pixels with a value of 1 represent anomaly regions. The dataset features a consistent file structure containing h5 files for different categories. Each hdf5 file includes images and corresponding masks, with images divided into anomaly and normal types, and masks providing pixel-level annotations for anomaly regions.