Training and Testing Data¶

Task 1: Segmentation of  Diabetic Retinopathy Lesions

Training set consists of 109 images and test set consists of 65 images. 

The dataset contains three different Diabetic Retinopathy Lesions: Intraretinal Microvascular Abnormalities (1), Nonperfusion Areas (2), Neovascularization(3).

Task 2: Image Quality Assessment

Training set consists of 665 images and test set consists of 438 images. 

The dataset contains three different image quality levels: Poor quality level (0), Good quality level (1), Excellent quality level (2).

Task 3: Diabetic Retinopathy Grading

Training set consists of 611 images and test set consists of 386 images.

The dataset contains three different diabetic retinopathy grades: Non-DR (0), NPDR (1), PDR (2).

NPDR: non-proliferatived diabetic retinopathy; PDR: proliferatived diabetic retinopathy.