# Image Segmentation

> AI gives each dot in a photo a label. Then you see the true shape of each thing.

**Field:** Artificial Intelligence  
**Category:** Robotics & Vision  
**Also known as:** Semantic Segmentation, Instance Segmentation, Pixel Segmentation

## Think of it like

It is like a coloring book. You color every part of the page by what it shows.

## Definition

Image segmentation is when an AI model gives each pixel in a picture a label. You get the exact shape of each thing, not just a box. One kind colors all pixels of one sort the same, so all cars are one color. Another kind also tells each car apart. It is used where the exact shape matters, such as a tumor in a scan, or cutting a person out of a photo.

## Example

A photo app cuts Lucy out of a selfie. It marks each pixel as Lucy or not Lucy, so the cut follows her hair.

## Why it matters

A box is rough. When the exact edge matters, as in a scan or a photo edit, you need each pixel.

## Common mistake

Using it when a box would do. Segmentation needs far more labeled data and computing power.

## Three principles

- **Every pixel.** Gets a label.
- **Exact shape.** Not a box.
- **Semantic or instance.** Kinds, or each one.

## Related terms

- [Object Detection](https://learn.benibix.com/term/ai/object-detection)
- [Image Classification](https://learn.benibix.com/term/ai/image-classification)
- [Computer Vision](https://learn.benibix.com/term/ai/computer-vision)
- [Inpainting](https://learn.benibix.com/term/ai/inpainting)
- [AI Background Removal](https://learn.benibix.com/term/ai/background-removal)

## Told apart from

- [Object Detection vs Image Segmentation](https://learn.benibix.com/compare/object-detection-vs-image-segmentation): People often think object detection and image segmentation give the same answer. Detection draws a box around each thing, and the box takes in some of the background too. Segmentation labels each pixel, so it gives the exact shape. A box is cheaper and faster. The exact shape is worth the cost when edges matter, as in a scan or a photo edit.

![The same street photo with every pixel colored by what it shows: road in one color, cars in another, people in a third.](https://learn.benibix.com/diagrams/rb-image-segmentation.svg)

---

Source: https://learn.benibix.com/term/ai/image-segmentation
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.
