LEAR's main focus of research is learning based approaches to visual object recognition and scene interpretation, particularly for image retrieval and video indexing. Understanding the content of everyday images and videos is one of the most challenging problems in computer vision. The extent to which we can do this is currently limited, but we believe that very significant advances will be made over the next few years by combining emerging statistical learning techniques with state of the art image descriptors. This field is also close to a major threshold of applicability: even partial solutions are likely to enable many new applications.