Hi, I’m building a hog-SVM detector with a focus on cars and pedestrians. Everything is in place however i read the following in the documentation for hog: link
“(II) no smoothing within cells (Gaussian spatial window with sigma=8pix in the paper)”
How do i go about to add this smoothing? I really want to implement the same solution as Dalal and Triggs in regards to person detection. As well as an implementation with cars, in where this smoothing is always a part of the detector.
I’m having a hard time grasping how i would implement this. Can someone please provide some information of how this alteration would be done?
Thanks a lot!