---
title: Visual Genome
type: templates
category: Computer Vision
cat: computer-vision
order: 155
meta_title: Label different described features
meta_description:
---
Use the [Visual Genome](https://visualgenome.org/) task to connect structured image concepts to language.
## Labeling Configuration
```xml
```
## Example data
```json
{ "data": {
"image": "https://htx-pub.s3.amazonaws.com/templates/visual-genome/panda.jpeg",
"regions": [
{ "value": "panda bear holding something in it's paw" },
{ "value": "trees behind the panda out of focus" },
{ "value": "piece of bamboo in panda's paw" },
{ "value": "bamboo next to panda" },
{ "value": "the panda is cute" }
],
"attributes": [
{ "value": "markings is black" },
{ "value": "nose is black" },
{ "value": "face is round" },
{ "value": "wood is round" }
],
"relationships": [
{ "value": "hole ON log" },
{ "value": "tree BEHIND bear" },
{ "value": "mouth OF panda" }
]
}}
```