How to get confidence values for detections from vision.Cas​cadeObject​Detector System object?

First, I run trainCascadeObjectDetector on a custom set of image data and save the result to an xml file.

Then, I create the vision.CascadeObjectDetector system object:

detector = vision.CascadeObjectDetector(xmlfile);

Finally, I want to run the detector on a test image:

bbox = step(detector,img);

Is there a way to compute confidence values for the detections returned here? For vision.PeopleDetector objects, you can run

 [bbox,scores] = step(detector,img);

but you cannot do this for vision.CascadeObjectDetector objects. Is there an easy workaround?

Answer 1

[Dima Lisin]

Hi Matt,

Unfortunately vision.CascadeObjectDetector does not return a confidence score, and there is no workaround. The reason vision.PeopleDetector does return a score, is because it is using a SVM classifier, which provides a score. vision.CascadeObjectDetector, on the other hand, uses a cascade of boosted decision trees, which does not lend itself well to computing a confidence score.

[Matthew Eicholtz]

Ah yes, this makes total sense. I had just transitioned from using SVM classifiers, so I suppose it slipped my mind that I was now using DTs instead. Thanks for the clarification.