Metrics used to evaluate an object detector’s performance.
- macOS 10.15+
- Xcode 11.0+
- Create ML
An object detector generates intersection-over-union (IoU) metrics, which is a way to measure the similarity of two bounding boxes. The IoU metric is the overlapping area divided by the area of the union of the bounding boxes.
For example, two bounding boxes that overlap perfectly have an IoU of
1, because their overlap is the same area as the union. Two bounding boxes that have no overlap have an IoU of
0. Anything between
1 either means the two bounding boxes partially overlap or one box completely encases the other.