dataset_info:
features:
- name: index
dtype: int64
- name: question
dtype: string
- name: A
dtype: string
- name: B
dtype: string
- name: C
dtype: string
- name: D
dtype: string
- name: E
dtype: string
- name: answer
dtype: string
- name: category
dtype: string
- name: clinical VQA task
dtype: string
- name: department
dtype: string
- name: perceptual granularity
dtype: string
- name: modality
dtype: string
- name: original task
dtype: string
- name: image_path
dtype: string
splits:
- name: validation
num_bytes: 72470811
num_examples: 10
download_size: 72466952
dataset_size: 72470811
task_categories:
- visual-question-answering
language:
- en
tags:
- medical
Modalities:
- image
- text
arxiv:
- https://arxiv.org/abs/2507.17539
license: creativeml-openrail-m
Fundus-MMBench
Benchmark for paper Constructing Ophthalmic MLLM for Positioning-diagnosis Collaboration Through Clinical Cognitive Chain Reasoning
🚨 Important: This benchmark is for academic research only.
Dataset Viewer Notice
🚨 The dataset viewer above only shows a preview of the first 10 rows from the dataset intending to provide a quick look at the data structure. The total number of data in the dataset is 620. To access the complete dataset, please download the full tsv file.
Introduction
The number of test samples for each task category of Fundus-MMBench is 20. It consists of 31 fine-grained tasks covering three core clinical domains: region-based object recognition (e.g., optic disc identification), disease classification (e.g., glaucoma vs. non-glaucoma diagnosis), and severity grading (e.g., diabetic retinopathy severity assessment).
Usage
You can run the evaluation on Fundus-MMBench using open-compass/VLMEvalKit. Note that Fundus-MMBench(tsv version) is not officially supported, but can be regarded as a Custom MCQ dataset.
Data Source
We would like to thank these contributions.