Our lab studies computer vision, whose final goal is to enable computers to perform the high-level visual information processing that human vision does. Towards this end, we study all sorts of inverse problems of extracting various types of information about objects/scenes from their images. Considering a wide range of research subjects from image sensing to artificial intelligence, we conduct both basic and applied researches. Examples of basic researches include accurately modeling the three-dimensional structures or its temporal changes of objects/scenes from their images and also the study of probabilistic graphical models and deep neural networks for visual recognition and image sensing. Examples of applied researches include the method of visually recognizing qualities of object surfaces, such as glossiness and roughness, as well as human vision, and the method of estimating spatiotemporal changes of a city from a large set of images, e.g., a collection of snapshots of the city taken by anonymous people. This research is motivated by Great East Japan Earthquake; we aim at visualizing the reconstruction process of the damaged cities over a long period of time. For this purpose, we have been periodically acquiring the images of the reconstruction processes of these areas by using a vehicle with a camera on its roof.