Curriculum Vitae

PDF version

Work experience

Postdoctoral Researcher, 2022 - 2024
- Hosted by GeomeriX, LIX, École Polytechnique, supervised by Maks Ovsjanikov.
- Ongoing project about end-to-end lightweight probabilistic hierarchical point cloud diffusion model.
- Ongoing project about a novel Neural Radiance Field (NeRF) representation suitable for 3D shape reconstruction and NeRF editing currently under review).
- VoroMesh: developed a deep learning model for a novel 3D shape representation, capable of producing watertight meshes from input SDF grids. Managed a PhD student over the course of the project.
- Self-supervised Dual Contouring: worked in a team with a PhD student on a project about a novel differentiable mesh reconstruction model based on Deep Learining and Dual Contouring. Explored probabilistic modifications of the proposed model.
Ph.D. Student, 2017 - 2021
- Hosted by Thoth and Morpheo, Inria Grenoble, supevised by Jakob Verbeek and Edmond Boyer.
- Discrete Point Flow Networks: developed a probabilistic generative model for point clouds based on invertible normalizing flows. Achieved comparable to state-of-the-art performance while reducing the computational load by magnitudes.
- Probabilistic Reconstruction Networks: developed a probabilistic framework for inference-based 3D shape reconstruction models from image inputs. Best science paper honorable mention at British Machine Vision Conference 2019.
Research Intern, 2016 - 2017
- Hosted by Skoltech Computer Vision Group, supervised by Victor Lempitsky.
- Research projects on CNN compression and 3D point cloud recognition.
- Escape From Cells: Kd-Network for 3D Shape Recognition: developed a novel feature extraction neural network model for classification, segmentation and retrieval tasks for point clouds based on construction of noisy kd-trees.

Education

Ph.D. in Mathematics and Informatics, Université Grenoble Alpes, 2021
M.S. in Data Science with Honors, Skolkovo Institute of Science and Technology, 2017
B.S. in Applied Mathematics and Physics, Moscow Institute of Physics and Technology, 2015

Additional Experience and Awards

Regular reviewer for CVPR, NeurIPS, ICLR, ICCV, ECCV, ICML, TPAMI, etc.
Organization of weekly laboratory seminars among the members of Maks Ovsjanikov team.
Best Science Paper Honorable Mention Award at British Machine Vision Conference, 2019.
Summer School PAISS 2018.
2nd place in ShapeNet Point Cloud Segmentation Challenge held at the International Conference on Computer Vision 2017.
Teaching for a summer school on introductory machine learning for high school students.
Organization of career fair days for university students in MIPT.

Skills, Activities, Interests

Scientific expertise: Computer Vision, Deep Learning, Probabilistic Modeling, 3D Shape Recognition, 3D Shape Generation, Surface Reconstruction, Differentiable 3D Shape Representations, Differentiable Rendering.
Recent technical experience: Python, PyTorch, Nerfstudio, HDF5, Numpy, Scipy.
Previous technical experience: TensorFlow, MongoDB, C++.
Languages: English (fluent), French (intermediate), Russian (native).
Hobbies: Running, Weight Lifting, Skiing, World of Warcraft competitive group activities (reached top 0.1% of players).

Publications

Self-Supervised Dual Contouring

R. Sundararaman, R. Klokov and M. Ovsjanikov. "Self-Supervised Dual Contouring." In CVPR'24.

VoroMesh: Learning Watertight Surface Meshes with Voronoi Diagrams

N. Maruani, R. Klokov, M. Ovsjanikov, P. Alliez and M. Desbrun. "VoroMesh: Learning Watertight Surface Meshes with Voronoi Diagrams." In ICCV'23.

Discrete Point Flow Networks for Efficient Point Cloud Generation

R. Klokov, E. Boyer and J. Verbeek. "Discrete Point Flow Networks for Efficient Point Cloud Generation." In ECCV'20.

Probabilistic Reconstruction Networks for 3D Shape Inference from a Single Image

R. Klokov, J. Verbeek and E. Boyer. "Probabilistic Reconstruction Networks for 3D Shape Inference from a Single Image." In BMVC'19.

Large-Scale 3D Shape Reconstruction and Segmentation from ShapeNet Core55

L. Yi, et at. "Large-Scale 3D Shape Reconstruction and Segmentation from ShapeNet Core55." In ICCVW'17.

Escape from Cells: Deep Kd-Networks for the Recognition of 3D Point Cloud Models

R. Klokov and V. Lempitsky. "Escape from cells: Deep kd-networks for the recognition of 3d point cloud models." In ICCV'17.

Roman Klokov