Efficient face detection using coarse sampling

Today, vast majority of the state-of-the art face detection methods involve scanning the input image using a sliding window and running a classifier on each of the windows. The biggest problem with this approach is the number of windows that need to be scanned. In this work, a new scanning method is proposed that is up to 6 times faster than the standard sliding window method. The proposed method involves performing a coarse search with a simple classifier to find the regions that have high probability of containing a face, and then performing a fine search only on these regions to find the exact locations of the faces.