README.md

---
annotations_creators:
- expert-generated
language_creators:
- expert-generated
language:
- en
license: gpl-3.0
multilinguality:
- monolingual
size_categories:
- 1K<n<10K
source_datasets:
- original
task_categories:
- multiple-choice
- question-answering
- visual-question-answering
task_ids:
- multiple-choice-qa
- visual-question-answering
- multi-class-classification
tags:
- multi-modal-qa
- figure-qa
- vqa
- scientific-figure
- geometry-diagram
- chart
- chemistry
dataset_info:
  features:
  - name: image_path
    dtype: string
  - name: question
    dtype: 'null'
  - name: answer
    dtype: string
  - name: prompt_reasoning
    dtype: 'null'
  - name: prompt_no_reasoning
    dtype: string
  - name: image_category
    dtype: string
  - name: task_category
    dtype: string
  - name: question_type
    dtype: string
  - name: response_options
    sequence: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  - name: decoded_image
    dtype: image
  splits:
  - name: syntheticgeometry__triangle
    num_bytes: 328198888.0
    num_examples: 10000
  - name: syntheticgeometry__quadrilateral
    num_bytes: 327409666.0
    num_examples: 10000
  - name: syntheticgeometry__length
    num_bytes: 411043854.0
    num_examples: 10000
  - name: syntheticgeometry__angle
    num_bytes: 397038300.0
    num_examples: 10000
  - name: syntheticgeometry__area
    num_bytes: 400289876.0
    num_examples: 10000
  - name: 3d__size
    num_bytes: 1930906822.0
    num_examples: 10000
  - name: 3d__angle
    num_bytes: 4093207706.0
    num_examples: 10000
  download_size: 7226264280
  dataset_size: 7888095112.0
configs:
- config_name: default
  data_files:
  - split: syntheticgeometry__triangle
    path: data/syntheticgeometry__triangle-*
  - split: syntheticgeometry__quadrilateral
    path: data/syntheticgeometry__quadrilateral-*
  - split: syntheticgeometry__length
    path: data/syntheticgeometry__length-*
  - split: syntheticgeometry__angle
    path: data/syntheticgeometry__angle-*
  - split: syntheticgeometry__area
    path: data/syntheticgeometry__area-*
  - split: 3d__size
    path: data/3d__size-*
  - split: 3d__angle
    path: data/3d__angle-*
---
# VisOnlyQA

This repository contains the code and data for the paper "[VisOnlyQA: Large Vision Language Models Still Struggle with Visual Perception of Geometric Information](https://arxiv.org/abs/2412.00947)".

VisOnlyQA is designed to evaluate the visual perception capability of large vision language models (LVLMs) on geometric information of scientific figures. The evaluation set includes 1,200 mlutiple choice questions in 12 visual perception tasks on 4 categories of scientific figures. We also provide a training dataset consisting of 70k instances.

* Datasets:
  * VisOnlyQA is available at [VLMEvalKit](https://github.com/open-compass/VLMEvalKit) 🔥🔥🔥
    * VisOnlyQA in VLMEvalKit is different from the original one. Refer to [this section](#vlmevalkit) for details.
  * Hugging Face
    * Eval-Real: [https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Real](https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Real)
    * Eval-Synthetic: [https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Synthetic](https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Synthetic)
    * Train: [https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Train](https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Train)
* Code: [https://github.com/psunlpgroup/VisOnlyQA](https://github.com/psunlpgroup/VisOnlyQA)

<p align="center">
<img src="readme_figures/accuracy_radar_chart.png" width="500">
</p>

```bibtex
@misc{kamoi2024visonlyqa,
    title={VisOnlyQA: Large Vision Language Models Still Struggle with Visual Perception of Geometric Information}, 
    author={Ryo Kamoi and Yusen Zhang and Sarkar Snigdha Sarathi Das and Ranran Haoran Zhang and Rui Zhang},
    year={2024},
    journal={arXiv preprint arXiv:2412.00947}
}
```

## Dataset

VisOnlyQA is provided in two formats: VLMEvalKit and Hugging Face Dataset. You can use either of them to evaluate your models and report the results in your papers. However, when you report the results, please explicitly mention which version of the dataset you used because the two versions are different.

### Examples

<p align="center">
<img src="readme_figures/examples.png" width="800">
</p>

### VLMEvalKit

[VLMEvalKit](https://github.com/open-compass/VLMEvalKit) provides one-command evaluation. However, VLMEvalKit is not designed to reproduce the results in the paper. We welcome using it to report the results on VisOnlyQA in your papers, but please explicitly mention that you used VLMEvalKit.

The major differences are:

* VisOnlyQA on VLMEvalKit does not include the `chemistry__shape_multi` split
* VLMEvalKit uses different prompts and postprocessing.

Refer to [this document](https://github.com/open-compass/VLMEvalKit/blob/main/docs/en/Quickstart.md) for the installation and setup of VLMEvalKit. After setting up the environment, you can evaluate any supported models on VisOnlyQA with the following command (this example is for InternVL2-26B).

```bash
python run.py --data VisOnlyQA-VLMEvalKit --model InternVL2-26B
```

### Hugging Face Dataset

The original VisOnlyQA dataset is provided in Hugging Face Dataset. If you want to reproduce the results in our paper, please use this version and code in the GitHub repository.

* Eval-Real: [https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Real](https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Real)
  * 500 instances for questions on figures in existing datasets (e.g., MathVista, MMMU, and CharXiv)
* Eval-Synthetic: [https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Synthetic](https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Eval_Synthetic)
  * 700 instances for questions on synthetic figures
* Train: [https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Train](https://huggingface.co/datasets/ryokamoi/VisOnlyQA_Train)
  * 70,000 instances for training (synthetic figures)

[dataset](https://github.com/psunlpgroup/VisOnlyQA/tree/main/dataset) folder of the GitHub repository includes identical datasets, except for the training data.

```python
from datasets import load_dataset

real_eval = load_dataset("ryokamoi/VisOnlyQA_Eval_Real")
real_synthetic = load_dataset("ryokamoi/VisOnlyQA_Eval_Synthetic")

# Splits
print(real_eval.keys())
# dict_keys(['geometry__triangle', 'geometry__quadrilateral', 'geometry__length', 'geometry__angle', 'geometry__area', 'geometry__diameter_radius', 'chemistry__shape_single', 'chemistry__shape_multi', 'charts__extraction', 'charts__intersection'])

print(real_synthetic.keys())
# dict_keys(['syntheticgeometry__triangle', 'syntheticgeometry__quadrilateral', 'syntheticgeometry__length', 'syntheticgeometry__angle', 'syntheticgeometry__area', '3d__size', '3d__angle'])

# Prompt
print(real_eval['geometry__triangle'][0]['prompt_no_reasoning'])
# There is no triangle ADP in the figure. True or False?

# A triangle is a polygon with three edges and three vertices, which are explicitly connected in the figure.

# Your response should only include the final answer (True, False). Do not include any reasoning or explanation in your response.

# Image
print(real_eval['geometry__triangle'][0]['decoded_image'])
# <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=103x165 at 0x7FB4F83236A0>

# Answer
print(real_eval['geometry__triangle'][0]['answer'])
# False
```

### Data Format

Each instance of VisOnlyQA dataset has the following attributes:

#### Features
* `decoded_image`: [PIL.Image] Input image
* `question`: [string] Question (without instruction)
* `prompt_reasoning`: [string] Prompt with intstruction to use chain-of-thought
* `prompt_no_reasoning`: [string] Prompt with intstruction **not** to use chain-of-thought
* `answer`: [string] Correct answer (e.g., `True`, `a`)

#### Metadata
* `image_path`: [string] Path to the image file
* `image_category`: [string] Category of the image (e.g., `geometry`, `chemistry`)
* `question_type`: [string] `single_answer` or `multiple answers`
* `task_category`: [string] Category of the task (e.g., `triangle`)
* `response_options`: [List[string]] Multiple choice options (e.g., `['True', 'False']`, `['a', 'b', 'c', 'd', 'e']`)
* `source`: [string] Source dataset
* `id`: [string] Unique ID

### Statistics

<p align="center">
<img src="readme_figures/stats.png" width="800">
</p>

## License

Please refer to [LICENSE.md](./LICENSE.md).

## Contact

If you have any questions, feel free to open an issue or reach out directly to [Ryo Kamoi](https://ryokamoi.github.io/) (ryokamoi@psu.edu).