Filip Ilic
PhD student at TUGraz
Computer Vision & Machine Learning

About


My interests are Computer Vision and Machine Learning. I investigated motion in videos — its modeling and importance — in deep video understanding models during my PhD with my supervisors Thomas Pock, and Richard Wildes. Additionally, I was also lucky enough to have wonderful collaborators and projects that allowed me to work on other topics such as denoising, generative modeling, and implicit neural representations.

Before that, I worked at CERN for the CMS experiment on performance critical monitoring and analysis software.

PhD at Vision, Learning & Optimization Group - ICG - TU Graz
Visiting Researcher at Center for Vision Research - York University - Toronto
Technical Student at CMS Experiment - CERN - Geneva

Publications


CVPR
Selective, Interpretable, and Motion Consistent Privacy Attribute Obfuscation for Action Recognition Filip Ilic, He Zhao, Thomas Pock, Richard Wildes 2024 Conference on Computer Vision and Pattern Recognition
GCPR
Score-Based Generative Models for Medical Image Segmentation using Signed Distance Functions Lea Bogensperger, Dominik Narnhofer, Filip Ilic, Thomas Pock 2023 German Conference on Computer Vision
WACV
Lightweight Video Denoising using Aggregated Shifted Window Attention Lydia Lindner, Alexander Effland, Filip Ilic, Thomas Pock, Erich Kobler 2023 Winter Conference on Applications of Computer Vision
ECCV
Is Appearance Free Action Recognition Possible? Filip Ilic, Thomas Pock, Richard Wildes 2022 European Conference on Computer Vision
OAGM
A study on robust feature representations for grain density estimates in austenitic steel Filip Ilic, Marc Masana, Lea Bogensperger, Harald Ganster, Thomas Pock 2021 Computer Vision and Pattern Analysis Across Domains
OAGM
On the Influence of Beta Cell Granule Counting for Classification in Type 1 Diabetes Lea Bogensperger, Marc Masana, Filip Ilic, Dagmar Kolb, Thomas R Pieber, Thomas Pock 2021 Computer Vision and Pattern Analysis Across Domains
WACV
Representing Objects in Video as Space-Time Volumes by Combining Top-Down and Bottom-Up Processes Filip Ilic, Axel Pinz 2020 Winter Conference on Applications of Computer Vision

Miscellaneous


Teaching
A visual introduction to Monte Carlo sampling A visual and theoretical introduction to Monte Carlo sampling methods, emphasizing their utility in approximating expectations and integrations where direct computation is challenging. We take a closer look at Importance, Rejection, and Inversion Sampling.
Physicalization
A DIY Galton Board Random Walks, Central Limits, and a sneaky Gaussian.
We’ll explore how dropping steel ball bearings into a board filled with pegs can visually simulate the binomial distribution and its gradual transformation into a normal distribution. Instructions on how to build your own are included!
Physicalization
The Analogue GIF A demo piece on the perception of temporal consistency.
The analogue GIF explores human perception of motion from static noise patterns, revealing that effective motion detection persists even with imperfect image alignment. It also shows how robust our visual processing abilities are.
Physicalization
Dense packing with spheres A model for grain formation in steel, and other packing problems.
Built with acrylic glass and steel ball bearings.
Nerd Alert
Remapping keys System wide key remapping on Linux and MacOS