Unsupervised Object Representation Learning using Translation and Rotation Group Equivariant V AE (Supplementary Material)