I have an implemented face detector of my own, but I faced a problem recently that OpenCV
solves with the minNeighbors
parameter for detectMultiScale()
function: I have a lot of false positives. What OpenCV does is leaves only rectangles that have a certain amount of rectangles nearby.
Faces usually have a lot of detections around them.
This is an example.
Intuitively, I understand how it works, but I want to know what algorithm is used under the hood, so that I can implement it myself.
Thank you.