---
license: apache-2.0

dataset_info:
  features:
  - name: image
    dtype: binary  # Binary image data
  - name: internal_id
    dtype: string
  - name: url
    dtype: string
  - name: annotation
    struct:
      - name: symmetry
        dtype: int64
      - name: richness
        dtype: int64
      - name: color aesthetic
        dtype: int64
      - name: detail realism
        dtype: int64
      - name: safety
        dtype: int64
      - name: body
        dtype: int64
      - name: lighting aesthetic
        dtype: int64
      - name: lighting distinction
        dtype: int64
      - name: background
        dtype: int64
      - name: emotion
        dtype: int64
      - name: main object
        dtype: int64
      - name: color brightness
        dtype: int64
      - name: face
        dtype: int64
      - name: hands
        dtype: int64
      - name: clarity
        dtype: int64
      - name: detail refinement
        dtype: int64
      - name: unsafe type # Not used for training
        dtype: int64
      - name: object pairing
        dtype: int64
  - name: meta_result
    dtype: sequence[int64] 
  - name: meta_mask
    dtype: sequence[int64]  
  splits:
  - name: train
    num_examples: 40743 
---

# VRDB-Image

This dataset contains aesthetic annotations for images. The annotations cover 18 aspects of visual aesthetics and quality assessment.


## Annotation Details

Each image in the dataset is annotated with the following attributes:

<table border="1" style="border-collapse: collapse;">
<tr>
<th>Dimension</th>
<th>Attribute</th>
</tr>
<tr>
<td rowspan="5">Composition</td>
<td>Symmetry</td>
</tr>
<tr>
<td>Object pairing</td>
</tr>
<tr>
<td>Main object</td>
</tr>
<tr>
<td>Richness</td>
</tr>
<tr>
<td>Background</td>
</tr>
<tr>
<td rowspan="5">Quality</td>
<td>Clarity</td>
</tr>
<tr>
<td>Color Brightness</td>
</tr>
<tr>
<td>Color Aesthetic</td>
</tr>
<tr>
<td>Lighting Distinction</td>
</tr>
<tr>
<td>Lighting Aesthetic</td>
</tr>
<tr>
<td rowspan="5">Fidelity</td>
<td>Detail realism</td>
</tr>
<tr>
<td>Detail refinement</td>
</tr>
<tr>
<td>Body</td>
</tr>
<tr>
<td>Face</td>
</tr>
<tr>
<td>Hands</td>
</tr>
<tr>
<td rowspan="2">Safety & Emotion</td>
<td>Emotion</td>
</tr>
<tr>
<td>Safety</td>
</tr>
</table>

### Example: Scene Richness (richness)
- **2:** Very rich  
- **1:** Rich  
- **0:** Normal  
- **-1:** Monotonous  
- **-2:** Empty  

For more detailed annotation guidelines, please refer to:
- [annotation_deatils](https://flame-spaghetti-eb9.notion.site/VisionReward-Image-Annotation-Details-196a0162280e80ef8359c38e9e41247e)
- [annotation_deatils_ch](https://flame-spaghetti-eb9.notion.site/VisionReward-Image-195a0162280e8044bcb4ec48d000409c)


## Additional Feature Details
The dataset includes two special features: `annotation` and `meta_result`.

### Annotation
The `annotation` feature contains scores across 18 different dimensions of image assessment, with each dimension having its own scoring criteria as detailed above.

### Meta Result
The `meta_result` feature transforms multi-choice questions into a series of binary judgments. For example, for the `richness` dimension:

| Score | Is the image very rich? | Is the image rich? | Is the image not monotonous? | Is the image not empty? |
|-------|------------------------|-------------------|---------------------------|----------------------|
| 2     | 1                      | 1                 | 1                         | 1                    |
| 1     | 0                      | 1                 | 1                         | 1                    |
| 0     | 0                      | 0                 | 1                         | 1                    |
| -1    | 0                      | 0                 | 0                         | 1                    |
| -2    | 0                      | 0                 | 0                         | 0                    |

Each element in the binary array represents a yes/no answer to a specific aspect of the assessment. For detailed questions corresponding to these binary judgments, please refer to the `meta_qa_en.txt` file.

### Meta Mask
The `meta_mask` feature is used for balanced sampling during model training:
- Elements with value 1 indicate that the corresponding binary judgment was used in training
- Elements with value 0 indicate that the corresponding binary judgment was ignored during training

## Data Processing

We provide `extract.py` for processing the dataset into JSONL format. The script can optionally extract the balanced positive/negative QA pairs used in VisionReward training by processing `meta_result` and `meta_mask` fields.

```bash
python extract.py [--save_imgs] [--process_qa]
```