Continuous Surface Embeddings
This work provides a more automated and generalizable method for establishing dense correspondences, which is a significant problem for robotics and computer vision applications dealing with deformable objects.
This paper tackles the problem of learning and representing dense correspondences in deformable object categories. The authors propose a new learnable image-based representation that predicts an embedding vector for each pixel, linking it to a 3D object mesh vertex. This approach performs on par with or better than state-of-the-art methods for human dense pose estimation and scales to new deformable animal categories.
In this work, we focus on the task of learning and representing dense correspondences in deformable object categories. While this problem has been considered before, solutions so far have been rather ad-hoc for specific object types (i.e., humans), often with significant manual work involved. However, scaling the geometry understanding to all objects in nature requires more automated approaches that can also express correspondences between related, but geometrically different objects. To this end, we propose a new, learnable image-based representation of dense correspondences. Our model predicts, for each pixel in a 2D image, an embedding vector of the corresponding vertex in the object mesh, therefore establishing dense correspondences between image pixels and 3D object geometry. We demonstrate that the proposed approach performs on par or better than the state-of-the-art methods for dense pose estimation for humans, while being conceptually simpler. We also collect a new in-the-wild dataset of dense correspondences for animal classes and demonstrate that our framework scales naturally to the new deformable object categories.