Skip to content
ml-recall

#precision-recall

4 questions from Object Detection, NLP Fundamentals

Timed test

4 questions · 4 min · score and explanations at the end

Take a test
Q 1 / 4Score 0 / 0

Question 1: Short answer

hardAI-checkedObject Detection

A detector is evaluated on one class with 4 ground-truth objects, GT1 to GT4. Its six detections, sorted by confidence, are listed with the ground-truth box each overlaps most and that IoU:

rank confidence best ground truth IoU
1 0.95 GT1 0.82
2 0.90 GT1 0.65
3 0.80 GT2 0.71
4 0.70 GT3 0.42
5 0.60 GT4 0.55
6 0.40 none 0

A detection is a true positive if its IoU is at least 0.5 and its ground truth has not already been matched by a higher-ranked detection; otherwise it is a false positive. AP is the sum of the precisions at the ranks of the true positives, divided by the number of ground-truth objects. What is AP?

a number between 0 and 1, 3 decimals