Textbook › Part II The Lineage of AI Models
CHAPTER 21
Detection and Segmentation
Everything so far has been classification: giving one answer to the question "what is in this image?" In practice you need something finer than that.
- Object detection
- What is there, where it is, and how many of them. The answer comes as rectangular frames (bounding boxes) drawn around the objects.
- Semantic segmentation
- Painting in every single pixel with the class it belongs to. This gives you things like the boundary between a road and a sidewalk.
- Instance segmentation
- Painting in the pixels while also telling individual objects of the same class apart. This gives you "the third person."
Two lineages
Object detection methods split into two broad lines as they developed.
Comments
Sign in to comment