Image Segmentation is a computer vision task that labels every pixel in an image, precisely outlining the exact shape of each object rather than just placing a box around it.
What it means in plain English
Where object detection draws rough boxes, segmentation is far more precise: it decides, pixel by pixel, which object each part of the image belongs to. This produces an exact outline of every object, which is essential when precise boundaries matter. It is the most detailed form of visual understanding.
The extra precision comes at the cost of more computation than simple detection.
A simple example
In medical imaging, segmentation outlines the exact boundary of a tumour on a scan, pixel by pixel, so its precise size and shape can be measured — something a rough bounding box could not provide.
Why it matters
Image segmentation is crucial wherever exact object boundaries matter — medical diagnosis, autonomous driving, photo editing tools that separate a subject from its background, and scientific analysis. It represents the most fine-grained way AI understands images.
Related terms
- Object Detection — the coarser task of boxing objects.
- Computer Vision — the field segmentation belongs to.
- Convolutional Neural Network — commonly used for segmentation.
Frequently asked questions
What is image segmentation?
It is dividing an image into regions and labelling each pixel, so the model identifies exactly which pixels belong to which object — more detailed than just drawing a box around an object.
Where is image segmentation used?
In medical imaging, self-driving cars, photo editing (like background removal), and any task needing precise, pixel-level understanding of an image.
