Image Segmentation
The short answer · 16 words
In short
Computer vision task in which image regions or individual pixels are assigned to classes or objects.
Flow-alerta.cab (2026): Image Segmentation. https://flow-alerta.cab/en/glossar/image-segmentation. Editorially reviewed on: 09 October 2026.
How much detail would you like?
Practical
Relevant if the shape, surface or exact position of visual structures is important, for example in quality control, medical images, satellite data or image processing.
Example scenario: On a production photo, the exact area of a visible material defect is marked pixel by pixel.
Use for visual processes in which not only the presence, but the exact spatial extent of an area is required.
At a glance
- Core
- In image segmentation, not only does the entire image receive a label, but pixels or regions are classified. Semantic segmentation assigns pixel classes; instance segmentation additionally separates individual object instances of the same class.
- Use
- Relevant if the shape, surface or exact position of visual structures is important, for example in quality control, medical images, satellite data or image processing.
- Example
- Example scenario: On a production photo, the exact area of a visible material defect is marked pixel by pixel.
- Limits
- Confuse segmentation with classification; do not distinguish semantic and instance segmentation; ignore poor mask quality in training; use pixel accuracy alone as a quality measure.
In context
Questions about this term
What is the difference with classification?
Segmentation provides spatial pixel information.
What is semantic segmentation?
Pixels are assigned to classes.
What is instance segmentation?
Individual objects of the same class are separated.
All details and background
In image segmentation, not only does the entire image receive a label, but pixels or regions are classified. Semantic segmentation assigns pixel classes; instance segmentation additionally separates individual object instances of the same class.
ELI5 – Explained simply
In a way, the model paints a mask over the image and marks which pixels belong to which area or object.
Why it matters in practice
Relevant if the shape, surface or exact position of visual structures is important, for example in quality control, medical images, satellite data or image processing.
Example
Example scenario: On a production photo, the exact area of a visible material defect is marked pixel by pixel.
Pitfalls and limits
- Confuse segmentation with classification; do not distinguish semantic and instance segmentation; ignore poor mask quality in training; use pixel accuracy alone as a quality measure.
Origin & history
Segmentation is an established subfield of computer vision. Modern deep networks like U-Net made pixel-accurate segmentation practically usable in many domains.
Use in mid-sized companies
- Use for visual processes in which not only the presence, but the exact spatial extent of an area is required.
Related terms
- Image classification
- Object recognition
- Computer vision
- U-Net
Matching Flow-alerta service
AI assistants & automationAI assistants & automation: test a suitable process with clear rights and human test points.Matching tool / use case
View use casesTool example: Keras with a U-Net-like architecture; use case: pixel-by-pixel allocation of image regions.Sources
- www.tensorflow.orgTensorFlow, Image segmentation
Quellenprüfung: 09.10.2026. Definition und dokumentierte Einordnung abgeglichen; Beispiele beschreiben mögliche Anwendungen.
Editorially reviewed on: