Evaluating metrics F1, F2, Mean Average Precision for object detection

Question

Up today in the company where I work we are using the F1 Score for evaluating the performance of our model, also our competitor's using the same metric.

1. I would like to understand what's the difference between F1,F2 and mAP?(Please do not explain me how to calculate them, also I know that F measure gives the same weight to the precision and recall while mAP choose the best precision from all recalls)

2. Why in competitions (e.g. PASCAL VOC) and articles for object detection I am reading it is always preferred to use mAP instead of F1 or F2 scores ?

Thanks !

Answer 1

AP is more accurate than the F scores because it considers the PR relation globally. Articles adopt mAP on VOC because it is the official metric and they have to do comparison with other methods which also adopt this metric. Other competetitons such as some text detections also adopt P R and F score as the default metrics.