Instance vs. Semantic
Semantic segmentation treats multiple objects of the same class as a single entity. In the example below, you can see that the labeler is responsible for annotating a single structure ("Vertebrae"), regardless of the vertebrae's classification (e.g. L1, S1, etc.).
On the other hand, instance segmentation treats multiple objects of the same class as distinct individual objects (or instances). In the example below, the annotator must create a unique annotation (as represented by an Instance in Redbrick AI) for each specific vertebrae.
On RedBrick AI, you can perform both semantic and instance segmentation by controlling how many instances of a particular category you create on the left side bar. If you create more than one instance of a single category, on export, you will be able to correlate the instance ID's in your segmentation mask to your category names.
Check out this video for an overview:
Last updated
