Learning 3D appearance models from video | IEEE Conference Publication | IEEE Xplore