What is the advantage of DaimlerPeopleDetector? [closed]

asked 2014-01-17 03:26:30 -0500

Flart gravatar image

updated 2014-01-17 04:09:56 -0500

What is the advantage of DaimlerPeopleDetector (cv::HOGDescriptor::getDaimlerPeopleDetector()) as compared with DefaultPeopleDetector?

Is it detector trained using training images of lower (than DefaultPeopleDetector) size?

It works faster, doesn't it?

edit retag flag offensive reopen merge delete

Closed for the following reason the question is answered, right answer was accepted by sturkmen
close date 2020-09-26 15:46:03.815009



Basically the DaimlerPeopleDetector is the support vector created by the research lab at Daimler which they used to perform pedestrian detection. It is created on another dataset than the defaultPeopleDetector. In occardance to performance, it all depends on your application.

StevenPuttemans gravatar imageStevenPuttemans ( 2014-01-17 04:14:42 -0500 )edit

Doesn't same application data and lower training image size implies increase in performance? Does quality of detection depend of my data, and no predetermination in question "what better" ?

Flart gravatar imageFlart ( 2014-01-17 05:03:28 -0500 )edit

not increase in performance, though lower training image size will lead to less features that need to be calculated in each window and thus giving a faster detector. However feature calculation is far from the slowest part of the algorithm. And yeah, your data actually ensures if stuff works well or not. These techniques all depend on the training data quality, like the variation contained, the features extracted, the amount of unique samples for the computer, ...

StevenPuttemans gravatar imageStevenPuttemans ( 2014-01-17 05:59:14 -0500 )edit