Glossary of Machine Learning Terms

Precision

Precision

Add an equation here

Recall

Recall

Add an equation here

F1

F1

IoU

Intersection-over-union. A metric describing the level of overlap between two shapes. It may be stated as a ratio (ranging from 0 to 1) or as a percentage (ranging from 0% to 100%).

The intersection is the .

An IoU of 50% is often used as a general .

mAP

Mean average precision. The apparent redundancy in this term (“Isn’t mean the same as average?”) causes some confusion, but this is just due to the fact that many models have multiple classes.

The average precision (AP) is the area under the curve of.

If you have a model that purely

mAP50

Mean average precision at an IoU of 50%.

mAP50-95

Mean average precision at a range of IoU thresholds, typically ranging