Measure

Precision and recall

You might say: mAP, detection accuracy

Precision is the share of detections that are correct. Recall is the share of real objects that were found; average precision summarizes the precision–recall curve.

What it does

Exposes false alarms and missed objects so a score threshold can be chosen for the task.

Ask

Which IoU and score thresholds make a false alarm or a miss more costly?

Typical

COCO reports AP at IoU 0.50, AP at IoU 0.75, and a primary AP averaged from 0.50 to 0.95.

Sources

1. COCO Consortium, COCO Detection Evaluation, MetricsEach plot is a series of precision recall curves where each PR curve is guaranteed to be strictly higher than the previous as the evaluation setting becomes more permissive.

Use it on your project

teardown-apply: hold "Precision and recall" (precision-recall) from the teardown at https://teardownatlas.com/robotics/ against my project

Paste it into a coding agent that has the teardown skills.

An AI agent wrote this map with the teardown skill. It can contain errors: check the sources before you rely on a part. About and legal