Name: Artem Sevastopolsky
Position: Ph.D Candidate
E-Mail: artem.sevastopolsky@gmail.com
Phone: +49-89-289-19486
Room No: 02.07.035

Bio

I am a Ph.D. student at the Visual Computing & Artificial Intelligence lab. Previously, I worked as a researcher at the Vision, Learning and Telepresence lab at Samsung AI Center Moscow with the focus on 3D deep learning, differentiable rendering, point cloud processing and human telepresence. My previous degrees were obtained from Skolkovo University of Science and Technology (M.Sc.) and Lomonosov Moscow State University (B.Sc.), both in machine learning and computer science. I have also worked in medical imaging during the last years of my B.Sc. and in the industry. My current research is related to face recognition, active learning, and neural rendering. Homepage

Research Interest

neural rendering, 3D vision, face generation, novel view synthesis

Publications

Preprints

HeadCraft: Modeling High-Detail Shape Variations for Animated 3DMMs
Artem Sevastopolsky, Philip-William Grassal, Simon Giebenhain, ShahRukh Athar, Luisa Verdoliva, Matthias Nießner
arXiv
We learn to generate large displacements for parametric head models, such as long hair, with high level of detail. The displacements can be added to an arbitrary head for animation and semantic editing.
[video][code][bibtex][project page]

2024

TriPlaneNet: An Encoder for EG3D Inversion
Ananta Bhattarai, Matthias Nießner, Artem Sevastopolsky
WACV 2024
EG3D is a powerful {z, camera}->image generative model, but inverting EG3D (finding a corresponding z for a given image) is not always trivial. We propose a fully-convolutional encoder for EG3D based on the observation that predicting both z code and tri-planes is beneficial. TriPlaneNet also works for videos and close to real time.
[video][code][bibtex][project page]

2023

How to Boost Face Recognition with StyleGAN?
Artem Sevastopolsky, Yury Malkov, Nikita Durasov, Luisa Verdoliva, Matthias Nießner
ICCV 2023
State-of-the-art face recognition systems require huge amounts of labeled training data which is often compiled as a limited collection of celebrities images. We learn how to leverage pretraining of StyleGAN and an encoder for it on large-scale collections of random face images. The procedure can be applied to various backbones and is the most helpful on limited data. We release the collected datasets AfricanFaceSet-5M and AsianFaceSet-3M and a new fairness-concerned testing benchmark RB-WebFace.
[video][code][bibtex][project page]

2020

TRANSPR: Transparency Ray-Accumulating Neural 3D Scene Point Renderer
Maria Kolos, Artem Sevastopolsky, Victor Lempitsky
3DV 2020
An extension of Neural Point-Based Graphics that can render transparent objects, both synthetic and captured in-the-wild.
[video][bibtex][project page]

Neural point-based graphics
Kara-Ali Aliev, Artem Sevastopolsky, Maria Kolos, Dmitry Ulyanov, Victor Lempitsky
ECCV 2020
Given RGB(D) images and a point cloud reconstruction of a scene, our neural network generates extreme novel views of the scene which look highly photoreal.
[video][code][bibtex][project page]

2019

Coordinate-based Texture Inpainting for Pose-Guided Image Generation
Artur Grigorev, Artem Sevastopolsky, Alexander Vakhitov, Victor Lempitsky
CVPR 2019
How would I look in a different pose? Or in different clothes? A ConvNet with coordinate-based texture inpainting to the rescue.
[code][bibtex][project page]