Neural network based face detection

One of the challenging problems in computer vision is face detection. The goal of face detection is to locate all regions that contain a face regardless of any three dimensional transformation and lighting condition. There are lots algorithms which were proposed for a solution of face detection problem. In this thesis, I have implemented a neural network based face detection approach. The system detects upright frontal faces in a given image. Some heuristics were used in order to reduce the number of false detections. Also more than one neural networks are used in order to improve the detection capability. There are lots of neural network based face detection systems in the literature. Since these proposed systems in the literature are complex they cannot be used for real time applications. First and the most important difference between the proposed system in this thesis and the systems in the literature is simplicity of a network. Another major difference is size of a window which is used for face detection. In this thesis 10×10 pixel size window which is the smallest window size in the literature is used for detection. Furthermore simple system and small size window gives a faster training and running.