Roman Klokov
Until recently I was a postdoctoral researcher working with Maks Ovsjanikov at École Polytechnique, Paris, France, where I was a member of the GeomeriX team. I recieved my PhD in Université Grenoble Alpes in December 2021. I was hosted by Inria Grenoble, supervised by Jakob Verbeek and Edmond Boyer and was a member of Thoth and Morpheo teams.
Broadly speaking, my research can be described as application of deep learning techniques to geometric 3D data, including multimodal data scenarios (geometry and appearance). I have worked on numerous applications in this domain, including differentiable rendering, differentiable shape meshing, generative modeling of 3D data in various representations, recognition of point clouds (classification, segmentation, retrieval). More precisely, I can underline three distinct areas of interest: 3D data representations suitable for deep learning and corresponding neural network architectures; probabilistic modeling for 3D data generation and inference-based reconstruction; neural differentiable applications for 3D data representations.
Currently I am open to consider any temporary or permanent working opportunities in France (hybrid/remote).