About
I'm a postdoc in the Computer Vision Group at TU Munich in Germany. My research focuses on visual scene understanding with limited supervision. Previously, I was a PhD student working at the Visual Inference Lab in TU Darmstadt. I obtained my Master's in Computer Science from University of Bonn. I also hold a Diploma degree in Mechanical Engineering from Moscow Aviation Institute (specialising in jets).
News
- 16/01/2024 Our work on direct image alignment has been accepted at ICLR '24 (oral presentation, top 1.2%).
- 20/10/2023 We will present a self-supervised approach for event data at WACV '24.
- 19/07/2023 Our work on self-adaptive semantic segmentation has been published at TMLR.
- 14/09/2022 I defended my PhD thesis "Deep Visual Parsing with Limited Supervision" with distinction (summa cum laude).
Teaching
Lecture |
Computer Vision 3: Segmentation, Detection and Tracking |
Practical Course |
Geometric Scene Understanding Course page (SS24, TBA) |
Publications
Flattening the Parent Bias:
|
|
An Analytical Solution to Gauss-Newton Loss for Direct Image AlignmentSergei Solonets*, Daniil Sinitsyn*, Lukas Von Stumberg, Nikita Araslanov and Daniel Cremers International Conference on Learning Representations (ICLR), 2024 (oral) To appear. |
|
Masked Event Modeling: Self-Supervised Pretraining for Event CamerasSimon Klenk*, David Bonello*, Lukas Koestler*, Nikita Araslanov and Daniel Cremers Winter Conference on Applications of Computer Vision (WACV), 2024 |
|
Semantic Self-adaptation:
|
|
Dense Unsupervised Learning for Video SegmentationNikita Araslanov, Simone Schaub-Meyer and Stefan Roth Advances in Neural Information Processing Systems (NeurIPS), 2021 Paper | Supplemental | Code |
|
Self-supervised Augmentation Consistency for Adapting Semantic SegmentationNikita Araslanov and Stefan Roth Conference on Computer Vision and Pattern Recognition (CVPR), 2021 Paper | Supplemental | Code |
|
Single-Stage Semantic Segmentation from Image LabelsNikita Araslanov and Stefan Roth Conference on Computer Vision and Pattern Recognition (CVPR), 2020 Paper | Supplemental | Code |