Antonin Vobecky
CTU in Prague
IMPACT team
Czech Technical University in Prague
Czech Institute of Informatics, Robotics and Cybernetics
Jugoslávských partyzánů 1580/3
160 00 Prague 6, Dejvice
Czech Republic
✉ antonin (dot) vobecky (at) cvut.cz
valeo.ai team
100 Rue de Courcelles, Paris
France
✉ antonin (dot) vobecky (at) valeo.com
I am a fourth year PhD student at the IMPACT (CTU/CIIRC) and valeo.ai teams working under the supervision of Josef Sivic (IMPACT), Patrick Pérez (Kyutai) and David Hurych (valeo.ai). Previously, I got a Master's degree and graduated with honors in Computer Science/Computer Vision from CTU in Prague. My research is in computer vision for scene understanding with a focus on applications in autonomous driving, namely weakly-supervised learning and multimodal learning (image, text, 3D and more).
News
07/2024 - A new paper "A study of test-time contrastive concepts for open-world, open-vocabulary semantic segmentation" (link) published on ArXiv! 🥳 06/2024 - Presenting OccFeat at CVPRw'24 04/2024 - OccFeat accepted to CVPR WAD workshop (paper) 🥳 01/2024 - MOCA accepted to the TMLR journal for publication 🥳 12/2023 - Presenting POP-3D at NeurIPS'23 👨🏫 09/2023 - POP-3D accepted to NeurIPS'23 🥳 07/2023 - ICVSS'23 essay competition winner 🏆 10/2022 - Presenting Drive&Segment at ECCV'22 🎙 07/2022 - Drive&Segment paper accepted as an ORAL to ECCV'22 conference 06/2022 - Scientific talks in INRIA WILLOW and at IMAGINE team at ENPC Paris. 03/2022 - Drive&Segment paper available on arXiv (full PDF) 02/2021 - DummyNet paper presented at AAAI'21. 10/2019 - Paper presented at ICCVw'19. 06/2019 - Master's degree at CTU in Prague, major Computer Vision. 09/2017 - Awarded Valeo Scholarship for talented students. 06/2017 - Bachelor's degree at CTU in Prague, major robotics, graduated with honours, Dean's award for bachelor thesis.
Publications
M. Wysoczanska, A. Vobecky, A. Cardiel, T. Trzcinski, R. Marlet, A. Bursuc, and O. SimeoniA study of test-time contrastive concepts for open-world, open-vocabulary semantic segmentation
ArXiv
[paper]
POP-3D: Open-Vocabulary 3D Occupancy Prediction from Images
NeurIPS'23
[Project page][paper][GitHub][Explanatory video]
MOCA: Self-supervised Representation Learning by Predicting Masked Online Codebook Assignments
TMLR
[paper]
Unsupervised Object Localization: Observing the Background to Discover Objects
CVPR'23
[Project page][paper][GitHub]
Drive&Segment: Unsupervised Semantic Segmentation of Urban Scenes via Cross-modal Distillation
ECCV'22 (oral).
[Project page][paper][arXiv][GitHub][Google Colab][Explanatory video]
Artificial Dummies for Urban Dataset Augmentation
In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2021.
[Project page][Paper on arXiv][GitHub]
Let's Get Dirty: GAN Based Data Augmentation for Camera Lens Soiling Detection in Autonomous Driving
In Proceedings of the Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2021.
[Paper]
Advanced Pedestrian Dataset Augmentation for Autonomous Driving
In Proceedings of the IEEE International Conference on Computer Vision Workshops (ICCVw), 2019.
[Paper]
Detecting decision ambiguity from facial images
In Proceedings of the IEEE International Conference on Automatic Face & Gesture Recognition (FG), 2019.
[Paper]
Desoiling dataset: Restoring soiled areas on automotive fisheye cameras
In Proceedings of the IEEE International Conference on Computer Vision Workshops (ICCVw), 2019.
[Paper]
Motion Prediction Influence on the Pedestrian Intention Estimation Near a Zebra Crossing
In Proceedings of the International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS), 2018.
[Paper]