Eye Diseases Classification

Dataset of retinal images classified into four categories: normal, diabetic retinopathy, cataracts and glaucoma. These images come from several open sources and are suitable for training medical imaging models.

Download dataset

Size

Approximately 4200 retinal images in image format (JPEG/PNG depending on source)

Licence

Open Database License (ODbL) with attribution to original authors

Description

‍

The dataset Eye Diseases Classification contains approximately 4200 wallpaper images divided into four main classes: normal, diabetic retinopathy, cataracts, and glaucoma. The images come from various open databases such as iDRID and HRF, offering visual diversity adapted to medical imaging research.

‍

What is this dataset for?

‍

Training models for the automatic classification of ocular diseases
Development of diagnostic tools in ophthalmology
Computer vision research applied to medical imaging

‍

Can it be enriched or improved?

‍

This dataset can be enriched by adding additional annotations specifying disease stages, or by integrating new images from other databases to improve diversity. Manual annotation by experts can improve the accuracy of classes.

‍

🔎 In summary

Criterion	Evaluation
🧩 Ease of use	⭐⭐⭐✩✩ (Standard format but metadata verification needed)
🧼 Need for cleaning	⭐⭐⭐⭐✩ (Low to moderate – format homogenization possible)
🏷️ Annotation richness	⭐⭐⭐✩✩ (Main classes present, detailed annotations limited)
📜 Commercial license	✅ Yes, with attribution required (ODbL)
👨‍💻 Beginner friendly	⚠️ Medium, medical imaging knowledge recommended
🔁 Fine-tuning ready	🖼️ Suitable for CNN fine-tuning on image classification
🌍 Cultural diversity	⚠️ Moderate – varied sources but focused on specific pathologies