FD3611: 3,611 Smartphone-Captured Retinal Images for Disease Classification

A collection of 3,611 color fundus photographs from the National Institute of Ophthalmology and Hospital in Dhaka, Bangladesh. The dataset was created by Kabir Ahmed Ridoy and last updated in April 2026 to support automated retinal disease diagnosis. Images were captured using a smartphone-assisted setup and are clinically validated by expert ophthalmologists.

Use Cases

• Training multi-class retinal disease classifiers based on expert-annotated diagnostic labels.
• Benchmarking model robustness based on real-world image variations in illumination, blur, and contrast.
• Developing AI diagnostic tools for low-resource settings based on smartphone-captured fundus images.

Strengths

• Contains 3,611 clinically validated images annotated by expert ophthalmologists.
• Preserves real-world variability in image quality, as mentioned in the description.
• Released under a permissive CC-BY-4.0 license for open research use.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.

Provenance

Source

National Institute of Ophthalmology and Hospital (NIOH), Dhaka, Bangladesh

Collection Method

Images were collected during routine ophthalmic examinations using a smartphone-assisted fundus imaging setup.

Freshness

Last updated 2026-04-25 16:44:52; freshness should be verified.

Geography

Dhaka, Bangladesh