In this report I present an approach for learning to identify human faces contained within greyscale images. Features are randomly extracted from a set of pre-training images that contain either informative or non-informative information about the object of interest. During this feature extraction process we vary either the total number of features that are extracted with a fixed window size, or we vary the window size of each feature. These features are then used to train three classification function. The trained functions will then be tested on images to predict if they contain a face or not, and these experiments are reported. We show that even with random features that may contain no information about the object, we can achieve high detection accuracies. Images used either contain a front view of a person's face, or an image of a object that is not a face.