profile photo

Ilia Chelak

EMAIL | RÉSUMÉ | LINKEDIN | GITHUB | X | SCHOLAR

Computer Vision PhD Student at UTokyo.

I am a doctoral student at UTokyo, co-supervised by Yoichi Sato (UTokyo) and Yasuyuki Matsushita (Microsoft Research). Previously, I was a research intern at Microsoft Research Asia.

My research interests lie at the intersection of embodied AI and 3D computer vision. I want to help robots truly understand the space around them, starting with knowing exactly where they are!

News

01/2026: I joined Microsoft Research Asia Tokyo! I will be working on pushing the frontiers of embodied AI research.
04/2025: I joined Prof. Yoichi Sato's lab at the University of Tokyo as a PhD student. I will be co-supervised by Prof. Yasuyuki Matsushita.
05/2024: I have graduated from the University of Helsinki! My thesis, centered on multi-view 3D reconstruction, sped up convergence by 4x (to 7 minutes down from 30) while greatly improving the geometry.
10/2023: I joined Matsushita group at Osaka University for a 6-month research exchange.

My Research

PriorSDF 3D reconstruction

PriorSDF: Fast and Accurate 3D Shape Recovery by SfM-informed Neural SDFs

Ilia Chelak

MSc Thesis, University of Helsinki 2024

Thesis | Bibtex 
@mastersthesis{
  title     = {{PriorSDF: Fast and Accurate 3D Shape Recovery by SfM-informed Neural SDFs}},
  author    = {Ilia Chelak},
  school    = {University of Helsinki},
  year      = {2024},
}
image description

Species-Agnostic Patterned Animal Re-identification by Aggregating Deep Local Features

Ekaterina Nepovinnykh, Ilia Chelak, Tuomas Eerola, Veikka Immonen, Heikki Kälviäinen, Maksim Kholiavchenko, Charles V. Stewart

IJCV 2024

Paper | Bibtex 
@article{
  author    = {Ekaterina Nepovinnykh and Ilia Chelak and Tuomas Eerola and Veikka Immonen and Heikki K{\"{a}}lvi{\"{a}}inen and Maksim Kholiavchenko and Charles V. Stewart},
  title     = {{Species-Agnostic Patterned Animal Re-identification by Aggregating Deep Local Features}},
  booktitle = {IJCV},
  year      = {2024}
}
image description

Matching individual Ladoga ringed seals across short-term image sequences

Ekaterina Nepovinnykh, Ilia Chelak, Andrei Lushpanov, Tuomas Eerola, Heikki Kälviäinen, Olga Chirkova

Mammalian Biology 2022

Paper | Bibtex 
@article{
  title     = {{Matching individual Ladoga ringed seals across short-term image sequences}},
  author    = {Ekaterina Nepovinnykh and Ilia Chelak and Andrei Lushpanov and Tuomas Eerola and Heikki K{\"{a}}lvi{\"{a}}inen and Olga Chirkova},
  booktitle = {Mammalian Biology},
  year      = {2022},
}
image description

EDEN: Deep Feature Distribution Pooling for Saimaa Ringed Seals Pattern Matching

Ilia Chelak, Ekaterina Nepovinnykh, Tuomas Eerola, Heikki Kälviäinen, Igor Belykh

Cyber-Physical Systems and Control II 2021 - Best paper first runner-up award

Paper | Bibtex 
@inproceedings{
  title     = {{EDEN: Deep Feature Distribution Pooling for Saimaa Ringed Seals Pattern Matching}},
  author    = {Ilia Chelak and Ekaterina Nepovinnykh and Tuomas Eerola and Heikki K{\"{a}}lvi{\"{a}}inen and Igor Belykh},
  booktitle = {Cyber-Physical Systems and Control II},
  year      = {2021},
}