Radim Špetlík is a PhD student in the Visual Recognition Group at the Czech Technical University, a co-founder and the chief technology officer of the start-up visionPigeon.ai (no longer active), the main developer and co-designer of the Mobile Game Xenofil (no longer available), and one of the three members of a team that placed first at the Angry Birds AI Competitions held on ECAI 2014 and IJCAI 2015, and third at IJCAI 2016.
Experience
Google LLC
Dec 2025 - Mar 2026Google Student Researcher (3 month internship)
Internship at XR research group in San Jose. Project related to Google XR.
Applied Computer Vision s.r.o.
Jul 2019 - PresentComputer Vision R&D Consultant
CNN-based detectors, predictors, and classifiers were designed and trained. Python web backends for CV applications developed. Neural networks quantized and deployed on ARM devices. Contract work for companies such as iC Systems.ai.
visionPigeon s.r.o.
Jun 2021 - Sep 2024Chief Technology Officer
Training and optimization of machine learning models for fashion e-commerce applications. Deployment on Azure cloud infrastructure. Design and implementation of annotation pipelines.
Microsoft Development Center Serbia
Jul 2018 - Sep 2018Computer Vision Research PhD intern
Improvement of the iris verification pipeline used in Windows Hello both in terms of accuracy and speed.
Selected Publications
Spetlik, R., Pliska, M., Vrba, V., Matas, J.
HelixTrack: Event-Based Tracking and RPM Estimation of Propeller-like Objects.
To appear in the Findings track, CVPR 2026.
#c++
Spetlik, R., Hlavsa, J., Čechová, J., Pojmanová, P., Matas, J., Urban, S.
Identity Verification from Human Scent using Channel Representation of 2D Gas Chromatography-Mass Spectrometry Data.
In Proceedings of Winter Conference on Applications of Computer Vision, 2026.
(github)
#python #highPerformanceComputing
Hlavsa, J., Spetlik, R., Čechová, J., Pojmanová, P., Matas, J., Urban, S.
Sex Classification from Human Scent Using Image Interpretation of 2D Gas Chromatography-Mass Spectrometry Data.
In Proceedings of Scandinavian Conference on Image Analysis, 2025.
(available online, github)
#python #highPerformanceComputing #leadership
Spetlik, R., Futschik, D., Sykora, D. (2025)
StructuReiser: A Structure-preserving Video Stylization Method.
In Computer Graphics Forum, 2025.
(available online, github)
#python #highPerformanceComputing
Spetlik, R., Matas, J. (2025)
Single-Image Localised Reflection Removal with k-Order Differences Term.
In Proceedings of Scandinavian Conference on Image Analysis, 2025.
(available online, github)
#python
Spetlik, R., Uhrova, T., Matas, J. (2025)
Efficient Real-Time Quadcopter Propeller Detection and Attribute Estimation with High-Resolution Event Camera.
In Proceedings of Scandinavian Conference on Image Analysis, 2025.
(available online, github)
#c++
Kolar, J., Spetlik, R, Matas, J. (2024)
EEPPR: Event-based Estimation of Periodic Phenomena Rate using Correlation in 3D.
In Proceedings of International Conference on Machine Vision, 2024.
(available online)
#leadership
Spetlik, R., Rozumnyi, D., Matas, J. (2024)
Single-Image Deblurring, Trajectory and Shape Recovery of Fast Moving Objects with Denoising Diffusion Probabilistic Models.
In Proceedings of Winter Conference on Applications of Computer Vision, 2024.
(available online)
#python #highPerformanceComputing
Spetlik, R., Razumenic, I. (2019)
Iris Verification with Convolutional Neural Network and Unit-Circle Layer.
In Proceedings of German Conference on Pattern Recognition, 2019.
(available online)
#python
Spetlik, R., Franc, V., Cech, J. and Matas, J. (2018)
Visual Heart Rate Estimation with Convolutional Neural Network.
In Proceedings of British Machine Vision Conference, 2018.
(available online)
#python #biometry
Research Projects
Computer Vision Methods for Analysis of Two-dimensional Gas Chromatography
Developing computer vision techniques to analyze and interpret GC×GC–MS data, focusing on tasks such as sex, or blood type classification. (Jan 2023 - Present)
Periodic Phenomena Attribute Estimation
Measuring key properties — most importantly frequency or rate — of repeating physical processes using high-temporal-resolution event data. (Jun 2024 - Present)
Temporally Consistent Style Transfer
Applying an artistic style to video such that the output frames are not only visually consistent with the chosen style, but also stable across time. (Jul 2023 - Present)
Single-image Fast Moving Object Recovery
Reconstructing the appearance and structure of an object that appears blurred or distorted in a photograph due to fast motion during exposure. (Jul 2021 - Present)
Single-image Reflection Removal
Separating the transmitted scene (what lies behind glass) from reflections (unwanted secondary layers) in a photograph taken through a reflective surface. (Feb 2020 - Jan 2021)
Non-invasive Estimation of Blood Glucose Level
Inferring glucose levels using optical signals that correlate with blood composition; a cooperation with Czech Institute of Endocrinology. (Jun 2019 - Present)
Skills
Education
-
PhD Degree2018 - Present, Czech Technical University in PragueSupervisor: prof. Ing. Jiří Matas, Ph.D
-
Master Degree (hons) in Computer Vision2016 - 2018, Czech Technical University in Prague
-
Bachelor Degree in Information Technology2013 - 2016, Czech Technical University in Prague
-
Bachelor Degree in General Studies & Humanities2010 - 2014, Charles University in Prague
Selected Awards
-
Rakathon - Hacker Community PrizeApr 2025
-
European Healthcare Hackathon - 2nd placeMar 2025
-
Chaos Neurathon - 1st placeNov 2023
-
IT SPY Best Master Thesis - Shortlisted2016 - 2018
-
Angry Birds AI Competition1st place (IJCAI 2015, ECAI 2014) • 3rd place (IJCAI 2016)