I have trained a cascade object detector in MatLab. On my test images, it is putting bounding boxes around multiple regions of interest (ROIs), whereas I only want to find the single best match. I.e: the ROI with the highest score. How can I either:
1) restrict the detector to one bounding box over the highest scoring ROI?
OR
2) extract the score data from the detector, so I can manually pick the best result?
I understand this is trivial using the built in HOG person detector, but I have not found a way of doing it using custom detectors.
Thanks.