Semantic Segmentation
Definition
Semantic segmentation assigns a content class to each image pixel.
A computer vision task that assigns a semantic class to each image pixel.
Simply explained
The system mentally colors all road surfaces in one color and all trees in another.
Why it matters in practice
The term helps to select, develop, evaluate AI systems in a well-founded manner and realistically assess their technical risks.
Example from business
A model marks pixels of component, background, and defect in a production image.
Limits and pitfalls
- Confusing semantic with instance segmentation or not evaluating edge quality.
Origin & history
Fully convolutional networks established modern semantic segmentation from 2015 onward.
Use in mid-sized companies
- Possible application: A model marks pixels of component, background, and defect in a production image.
Frequently asked questions
What does Semantic Segmentation mean?
A computer vision task that assigns a semantic class to each image pixel. What mistakes should be avoided? Confusing semantic with instance segmentation or not evaluating edge quality.
Sources
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: