Lesson 4 · Computer Vision
Segmentation
7 min
You'll be able to
- Understand pixel-level segmentation
- Distinguish semantic from instance segmentation
- See segmentation use cases
Segmentation labels every pixel in an image. Semantic segmentation groups pixels by class (all 'road' pixels), while instance segmentation separates individual objects (car A vs. car B).
Segmentation is essential for medical imaging — outlining tumours — and for autonomous driving, where precise pixel boundaries help vehicles understand their environment.
Challenge
Medical imaging
Explain why pixel-level segmentation (rather than a bounding box) might be preferred for outlining a tumour in an MRI scan.
Knowledge Check
Segmentation
Semantic segmentation labels:
Instance segmentation separates individual objects of the same class.
Answer all questions to submit.