---
license: cc-by-nc-4.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
dataset_info:
  features:
  - name: filename
    dtype: string
  - name: id
    dtype: int64
  - name: image
    dtype: image
  - name: source
    dtype: string
  - name: height
    dtype: int64
  - name: width
    dtype: int64
  - name: image_type
    dtype: string
  - name: contains_eye
    dtype: bool
  - name: contains_nose
    dtype: bool
  splits:
  - name: train
    num_bytes: 4563736373
    num_examples: 60669
  download_size: 4846624699
  dataset_size: 4563736373
task_categories:
- image-classification
- image-segmentation
- image-to-text
language:
- en
tags:
- dental
size_categories:
- 10K<n<100K
---

<h1 align="center">MetaDent: Labeling Clinical Images for Vision-Language Models in Dentistry</h1>

<p align="center">
  <a href="https://journals.sagepub.com/doi/10.1177/00220345261424242">
    <img src="https://img.shields.io/badge/Paper-JDR_2026-b31b1b" alt="Paper">
  </a>
  <a href="https://menxli.github.io/metadent/">
    <img src="https://img.shields.io/badge/Website-MetaDent-skyblue" alt="Website">
  </a>
</p>


### Overview

**MetaDent Image** is a large-scale clinical dental image dataset. It is constructed by collecting and filtering images from multiple publicly available and institutional sources. The dataset is designed to support research in dental image analysis and vision-language modeling for oral healthcare.

### Data Sources

The dataset is curated from the following sources:

* DS1: Hospital of Stomatology, Wuhan University (n=4,373)
* DS2: [Teeth or Dental Image Dataset](https://data.mendeley.com/datasets/6zsnhrds9t/1) (n=9,390)
* DS3: [COYO-700M](https://huggingface.co/datasets/kakaobrain/coyo-700m) filtered for dental images (n=46,906)

### More Information

For additional details about the dataset, please visit our project website:

👉 [https://menxli.github.io/metadent/](https://menxli.github.io/metadent/)