# Visual Question Answering

> You ask about a photo, and AI answers you in words.

**Field:** Artificial Intelligence  
**Category:** Multimodal  
**Also known as:** VQA, Image Question Answering

## Think of it like

It is like asking a friend about a photo. How many cats? What color is the door?

## Definition

Visual question answering is when an AI model answers a question about a picture. You give it a photo and a question, and it replies in words. You can ask what is there, how many, what color, what is written, or what is going on. The answer depends on what you ask, so one photo can give many answers. That is how it differs from a caption. Today this is mostly done by vision language models.

## Example

Jack takes a photo of his fridge and asks how many eggs are left. The model looks at the shelf and answers four.

## Why it matters

It gives you the one fact you need from a photo, not a long account of all of it.

## Common mistake

Asking about tiny things in a blurry photo. If you cannot see it well, the model may not see it either.

## Three principles

- **Picture plus question.** In words.
- **Answer in words.** Short and direct.
- **Depends on the ask.** One photo, many answers.

## Related terms

- [Vision Language Model (VLM)](https://learn.benibix.com/term/ai/vision-language-model)
- [Image Captioning](https://learn.benibix.com/term/ai/image-captioning)
- [Multimodal Model](https://learn.benibix.com/term/ai/multimodal-model)
- [Question Answering](https://learn.benibix.com/term/ai/question-answering)

## Told apart from

- [Image Captioning vs Visual Question Answering](https://learn.benibix.com/compare/image-captioning-vs-visual-question-answering): People think captioning and visual question answering are the same job. They are not. A caption is one fixed line that says what is in a picture, the same for everyone. Visual question answering waits for your question, and one photo gets a new answer for each one.

![A photo of a fridge shelf and the question how many eggs are left go into a model together. The model answers four.](https://learn.benibix.com/diagrams/mm-visual-question-answering.svg)

---

Source: https://learn.benibix.com/term/ai/visual-question-answering
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-17
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
