Which metrics of COCO API are suitable for object detection

0

These days I train a person detector and I use COCO API to evaluate my model. It uses Recall and Precision to evaluate a detection task.

It's output is something like this:

 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.505
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.697
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.573
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.586
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.519
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.501
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.387
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.594
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.595
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.640
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.566
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.564

In each epoch model performance change of validation test. Therefore I want to save my model each time that an improvement occurs.

My question is COCO-API gives 12 different average precision and average recall how should and select a model as better model in respect to other? Should I pay attention to AP or AR?

Farhad Bat

Posted 2019-10-14T15:16:53.630

Reputation: 105

No answers