FairVLMed: Fair Vision-Language Medical Ophthalmic Dataset

10,000 scanning-laser ophthalmoscopy fundus images paired with de-identified clinical notes, visual-field measurements, glaucoma labels, and demographic attributes.

At a glance

Field	Value
Short name	fairvlmed
Full name	FairVLMed: Fair Vision-Language Medical Ophthalmic Dataset
Publication date	Unknown
Date basis	Unknown
Publication date precision	Unknown
Publication date evidence	Unknown
Publication date source field	Unknown
Publication date reviewed	Unknown
Date notes	-
Primary category	multimodal
Resource role	current_dataset
Dataset family	fairvlmed
Contained modalities	fundus, visual_field, text, tabular
Tasks	classification, report_generation, fairness_analysis
Primary reported quantity	10,000 images
Classes	Not reported (Not reported)
Splits	train, val, test
Size	10.0 GB
Source-stated terms	CC BY-NC-ND 4.0
Normalized terms	cc-by-nc-nd
Descriptive screening label	Explicit noncommercial clause recorded; check source
Terms scope	dataset_files
Access friction	self_service_authenticated
Route backend	HuggingFace Hub
Availability	available (checked 2026-07-21)
Acquisition support	standard_platform_supported
Legacy sample-loader status	Metadata and access only

Reported quantities

Role	Count	Unit	Scope	Basis	Evidence
Primary	10,000	images	Primary quantity reported in the reviewed catalog source	legacy_catalog_field	huggingface.co/datasets

Counts retain their source-reported units. Additional rows can describe components, paired items, or derivative copies and are not automatically added to the primary quantity.

Notes

The official dataset card reports 10,000 patients and 10,000 samples (7,000 train, 1,000 validation, 2,000 test). No source statement supporting cohort identity with the separate Harvard-FairVision record was found, so no catalog relationship is asserted.

Access information and download

CLI

# Read-only preflight
eyehub download fairvlmed --data-dir ./data --dry-run --json

# Download, only when preflight reports supported behavior
eyehub download fairvlmed --data-dir ./data

Python

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

ds = REGISTRY.get_dataset('fairvlmed')
print(preflight_dataset(ds, './data'))  # no download

Upstream page: huggingface.co/datasets

Source-term evidence: huggingface.co/datasets

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{fairvlmed,
  title  = { FairVLMed: Fair Vision-Language Medical Ophthalmic Dataset },
  note   = { Harvard AI Robotics, FairVLMed: Fair vision-language medical ophthalmic dataset. HuggingFace, 2024 },
  year   = { 2024 },
  url    = { https://huggingface.co/datasets/harvardairobotics/FairVLMed },
}

Plain text

Harvard AI Robotics, FairVLMed: Fair vision-language medical ophthalmic dataset. HuggingFace, 2024.

Source-stated terms

• Raw source string: CC BY-NC-ND 4.0
• Normalized category: cc-by-nc-nd
• Apparent scope: dataset_files
• Descriptive screening label: Explicit noncommercial clause 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.

Similar resources by shared modality

• lmod_plus: LMOD+ Multimodal Ophthalmology Benchmark (32,633 annotated instances, unknown)
• grape: GRAPE: Glaucoma Real-world Appraisal Progression Ensemble (1,115 examinations, cc0)
• ocular_chat_vqa: OcularChat-VQA: AREDS-Derived Patient-Physician Dialogue Dataset (844,000 records, cc-by-nc-sa)
• eyecare_100k: Eyecare-100K: Multimodal Ophthalmology VQA Corpus (102,000 question answer pairs, unknown)
• angioreport: AngioReport Fundus Angiography Report Dataset (55,361 images, unknown)
• ffa_ir: FFA-IR Medical Report Dataset (47,247 images, unknown)
• x_pcr: X-PCR Ophthalmology Progressive Clinical Reasoning Benchmark (18,735 rows, unknown)
• deepeyenet: DeepEyeNet (DEN): Fundus Report Generation Dataset (15,709 images, research-only)