Are face recognition methods useful for classifying ships?

Face recognition research has gained significant interest in recent years which has resulted in the development of many state-of-the-art methods. However, it is not well-known how domain specific these methods are to the problem of face recognition. Could these algorithms be used to classify and identify other objects, such as ships seen from electro-optical satellite imagery? Face recognition research shares many of the same challenges with many other types of classification research, such as illumination, pose and resolution variation. Therefore, a study of this type is warranted. We present a comparison of several classical (e.g. eigenfaces and fisherfaces) as well as recent state-of-the-art face recognition methods (e.g. sparse representation and local binary patterns) using one standard face database and two databases of ship images collected from satellite imagery. An analysis of these results as well as future directions conclude the paper.