News

In contrast, we propose an innovative 3D Modulated Morphable Model (3D3M) to learn the dense shape correspondence from monocular images in a self-supervised manner. Specifically, given a batch of ...