OpenCV 4 with Python Blueprints - Second Edition

Precision

Precision in binary classification is a useful metric for measuring the fraction of retrieved instances that are relevant (also called the positive predictive value). In a classification task, the number of true positives is defined as the number of items correctly labeled as belonging to the positive class.

Precision is defined as the number of true positives divided by the total number of positives. In other words, out of all the pictures in the test set that a classifier thinks to contain a cat, precision is the fraction of pictures that actually do contain a cat.

Note that here, we have a positive label; thus, precision is a per-class value. We usually talk about the precision of one class or the precision of cats, and so on. ...