DeepEyeNet (DEN): Fundus Report Generation Dataset

15,709 fundus images with paired medical reports and extracted keywords. Only public fundus report-generation dataset — useful for VLM / captioning evaluation.

At a glance

Field	Value
Short name	deepeyenet
Full name	DeepEyeNet (DEN): Fundus Report Generation Dataset
Primary category	fundus
Contained modalities	fundus, text
Tasks	classification, multilabel
Samples	15,709
Classes	Not reported (Not reported)
Splits	train, val, test
Size	5.0 GB
Source-stated terms	Research only (NDA via email request)
Normalized terms	research-only
Descriptive screening label	Research or challenge restriction recorded; check source
Terms scope	dataset_files
Access friction	author_contact
Route backend	Manual (upstream-gated)
Availability	available (checked 2026-07-21)
Acquisition support	manual_access_blocked
Legacy sample-loader status	Metadata and access only

Notes

NDA gated — email [email protected] to request. No automated mirror exists.

Access preflight and acquisition

CLI

# This route requires upstream human action; no transfer starts.
eyehub download deepeyenet --data-dir ./data --dry-run --json
# Follow the official instructions shown by preflight.

Python

from eyedatahub.acquisition import preflight_dataset
from eyedatahub.datasets.registry import REGISTRY

ds = REGISTRY.get_dataset('deepeyenet')
print(preflight_dataset(ds, './data'))  # returns manual_access_blocked

Upstream page: github.com/Jhhuangkay

Source-term evidence: github.com/Jhhuangkay

Loader status

This catalog record provides metadata and access instructions, but it does not yet include a standard DatasetSample loader. Inspect the source file structure or contribute a loader before using it in a training pipeline.

Citation

BibTeX

@misc{deepeyenet,
  title  = { DeepEyeNet (DEN): Fundus Report Generation Dataset },
  note   = { Huang et al., 'DeepOpht: Medical Report Generation for Retinal Images via Deep Models and Visual Explanation', WACV 2021 },
  year   = { 2021 },
  url    = { https://github.com/Jhhuangkay/DeepOpht-Medical-Report-Generation-for-Retinal-Images-via-Deep-Models-and-Visual-Explanation },
}

Plain text

Huang et al., 'DeepOpht: Medical Report Generation for Retinal Images via Deep Models and Visual Explanation', WACV 2021.

Source-stated terms

• Raw source string: Research only (NDA via email request)
• Normalized category: research-only
• Apparent scope: dataset_files
• Descriptive screening label: Research or challenge restriction recorded; check source

⚠️ Source-stated terms, scope, and normalized labels are curation metadata, not legal advice or a permission finding. Review the current official source before transfer or reuse.

Related datasets with shared modalities

• eyecare_100k: Eyecare-100K: Multimodal Ophthalmology VQA Corpus (102,000 records, unknown)
• angioreport: AngioReport Fundus Angiography Report Dataset (55,361 records, unknown)
• ffa_ir: FFA-IR Medical Report Dataset (47,247 records, unknown)
• lmod_plus: LMOD+ Multimodal Ophthalmology Benchmark (32,633 records, unknown)
• x_pcr: X-PCR Ophthalmology Progressive Clinical Reasoning Benchmark (18,700 records, unknown)
• fundus_report_dataset: Fundus Report Dataset (422 records, cc-by)
• csdi: CSDI: Cataract Severity Diagnostic Image Dataset (187 records, cc-by)
• dme_vqa: Diabetic Macular Edema Visual Question Answering Dataset (count not reported records, cc-by)