---
license: cc-by-nc-4.0
language:
- en
- bg
- zh
- hr
- it
- sr
---
---
license: cc-by-nc-4.0
language:
- en
- bg
- zh
- hr
- it
- sr
---
# ImageCLEF 2026 Multimodal reasoning Open-Ended QA Dataset - Visual questions

## Introduction
The training data in this dataset are re-purposed from the [MBZUAI-IFM/EXAMS-V](https://huggingface.co/datasets/MBZUAI-IFM/EXAMS-V) dataset. Specifically, the original Multiple-Choice Questions (MCQs) have been converted into Open-Ended QA tasks to challenge models with free-form answer generation. 

In order to participate, please register on the evaluation platform: [AI4Media-Bench OpenQA Visual](https://ai4media-bench.aimultimedialab.ro/competitions/15/)

## Dataset Distribution 

### Language Distribution
The dataset is distributed across several languages as follows:

| Language | Train | Dev | Test | **Total** |
| :--- | :---: | :---: | :---: | :---: | 
| Bulgarian | 84 | 50 | 75 | **209** |
| Chinese | 139 | 50 | 75 | **264** |
| Croatian | 96 | 50 | 75 | **221** |
| English | 66 | 50 | 75 | **191** |
| Italian | 87 | 20 | 74 | **181** |
| Serbian | 56 | 20 | 65 | **141** |
| **Total** | **528** | **240** | **439** | **1207** |

### Graphical Distribution
The image-based portion of the dataset (`openqa-image`) contains various graphical elements:

| Graphical Feature | Train | Dev | Test | **Total** | Percentage (Overall) |
| :--- | :---: | :---: | :---: | :---: | :---: |
| Diagrams | 241 | 149 | 235 | **625** | 51.78% |
| Graphs | 107 | 44 | 98 | **249** | 20.63% |
| Tables | 92 | 35 | 71 | **198** | 16.40% |
| Other Figures | 88 | 35 | 96 | **219** | 18.14% |