Leon Sick

Hi and welcome! I’m currently a Ph.D. student at Ulm University where I’m part of the Visual Computing Group. My main supervisor is Timo Ropinski, but I also work closely with Pedro Hermosilla. In addition, I’m currently a Student Researcher at Google in Zürich, where I’m part of the Semantic Perception Team of Federico Tombari. My research centers on label-efficient and self-supervised learning, from unsupervised image and video segmentation to self-supervised representation learning. More recently, I’m especially interested in data-centric post-training of foundation models, in particular synthetic data generation, model-based data filtering and quality estimation, and self-improvement loops for LLMs and VLMs that require minimal human supervision.

And when I’m not researching, you can either find me on a sail boat or fishing during summer, but also skiing in the winter.

news

Sep 24, 2026	Thrilled to share that our work on S2D, an unsupervised video instance segmentation model trained on real video, is accepted at NeurIPS 2026! Congrats to Lukas Hoyer, Dominik Engel, Pedro Hermosilla and Timo Ropinski!
Oct 29, 2025	I’m happy to be named top reviewer for a second time at NeurIPS 2025! I also just attended the ICCV2025 Doctoral Consortium, it was a pleasure to have been mentored by Eric Brachmann.
Jun 30, 2025	Aloha!! Our work on CutS3D for Unsupervised Instance Segmentation is accepted at ICCV 2025! Congrats to all my amazing co-authors!
Mar 10, 2025	Very happy to have our work on Masked Scene Modeling as a 3D-native self-supervised pre-training framework accepted at CVPR 2025! Congrats to Pedro Hermosilla!
Jan 21, 2025	Super excited that our work on attention-guided masked autoencoders was awarded an Oral presentation at WACV 2025!

publication highlights

2026

S2D: Sparse-To-Dense Keymask Distillation for Unsupervised Video Instance Segmentation

Leon Sick, Lukas Hoyer, Dominik Engel, Pedro Hermosilla, and Timo Ropinski

In Advances in Neural Information Processing Systems (NeurIPS), Dec 2026

arXiv HTML

2025

CutS3D: Cutting Semantics in 3D for 2D Unsupervised Instance Segmentation

Leon Sick, Dominik Engel, Sebastian Hartwig, Pedro Hermosilla, and Timo Ropinski

In International Conference on Computer Vision (ICCV), Oct 2025

arXiv HTML
Masked Scene Modeling: Narrowing the Gap Between Supervised and Self-Supervised Learning in 3D Scene Understanding

Pedro Hermosilla, Christian Stippel, and Leon Sick

Conference on Computer Vision and Pattern Recognition (CVPR), Oct 2025

arXiv HTML
Attention-Guided Masked Autoencoders for Learning Image Representations

Leon Sick, Dominik Engel, Pedro Hermosilla, and Timo Ropinski

In Oral Presentation, Proceedings of the Winter Conference on Applications of Computer Vision (WACV), Feb 2025

arXiv HTML

2024

Unsupervised Semantic Segmentation Through Depth-Guided Feature Correlation and Sampling

Leon Sick, Dominik Engel, Pedro Hermosilla, and Timo Ropinski

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2024

arXiv HTML
TTT-KD: Test-Time Training for 3D Semantic Segmentation through Knowledge Distillation from Foundation Models

Lisa Weijler, Muhammad Jehanzeb Mirza, Leon Sick, Can Ekkazan, and Pedro Hermosilla

International Conference on 3D Vision (3DV), Jun 2024

arXiv
Weakly Supervised Virus Capsid Detection with Image-Level Annotations in Electron Microscopy Images

Hannah Kniesel, Leon Sick, Tristan Payer, Tim Bergner, Kavitha Sshaga Devan, and 4 more authors

International Conference on Learning Representations (ICLR), Jun 2024

HTML

2023

Leveraging Self-Supervised Vision Transformers for Neural Transfer Function Design

Dominik Engel, Leon Sick, and Timo Ropinski

Transactions On Visualization and Computer Graphics (TVCG), Jun 2023

arXiv