AIROGS

该数据集名为Rotterdam EyePACS AIROGS，包含了113,893张彩色眼底图像。其中，只有训练集是公开的，可供下载，包含101,442张可评估的图像。测试集则包含11,000张可评估和不可评估的图像，但不向公众开放。值得注意的是，该数据集存在明显的类别不平衡问题，'无参照性青光眼'类别的数量大约是'referable glaucoma'类别的15倍。数据集的分辨率较高，图像大小不一。该数据集的任务是对青光眼进行分类。

The dataset is named Rotterdam EyePACS AIROGS, which contains 113,893 color fundus images. Only the training set is publicly available for download, including 101,442 evaluable images. The test set consists of 11,000 images, both evaluable and non-evaluable, but it is not accessible to the general public. Notably, this dataset exhibits a significant class imbalance: the number of samples in the 'non-referable glaucoma' category is approximately 15 times that of the 'referable glaucoma' category. The dataset has high-resolution images with varying sizes, and its core task is glaucoma classification.