This project involves using OpenCV's trained Face Detection Neural Network to detect faces in various images and extract them either as separate images or create boxes around faces.
I am trying the dnn face detector from opencv. Did you try taking a photo close to the camera?. I did this, and found more than one detection ( more than one face) and some of the bounding box coordinates are outside the image and they are not faces ( i mean, i.e if the images is 640480 some bboxes coordinates are 720680 from the top left corner)