viola jones algorithm clarification

Question

As mentioned in the paper, while training the classifier 24x24 images are used.

But for face detection on image of size 1024x1600 which contains face of size 500x500 how is the face detected if we are using 24x24 windows to detect face?

Answer 1

The algorithm disassembles the image into 24-by-24 sub-images and in each sub-image, he tries to find a face. Due to this fact, the viola jones algorithm can find more than one face in the image.

I find a really good summary of the viola jones algorithm process https://drive.google.com/file/d/1Sxfp5k249eoxB57cbcS9vl4zi3gBe0la/view

suumery

you can read more about that here Image Analysis: 15th Scandinavian Conference, SCIA 2007, Aalborg, Denmark