My research interests focus on visual localization, 3D/4D reconstruction, and neural scene representations and understanding. I am also broadly interested in object-level perception, egocentric vision, and developing algorithms for robotics and mixed reality applications. Please feel free to drop me an email if you would like to discuss research ideas.

Recent deep learning have boosted the performance of visual SLAM. Methods like DROID-SLAM and DPVO rely on large-scale synthetic data to learn correspondences and camera poses, but they require ground-truth supervision, which limits the scale of available training data. In this project, we explore reducing the supervision signal to eventually develop a self-supervised visual SLAM system trainable on any casual video.

Preferred Requirements:

• Strong research motivation with a focus on achieving publication.
• Strong background in 3D computer vision and multi-view geometry; knowledge of SLAM is a plus.
• Proficiency in Python and PyTorch.

If you are interested, please contact me via email with your CV and transcript.

Export as PDF, XML, TEX or BIB

Journal Articles
2023
[]		Uncertainty-driven dense two-view structure from motion (W Chen, S Kumar and F Yu), In IEEE Robotics and Automation Letters (RA-L) and IROS 2023, IEEE, volume 8, 2023.  [bibtex] [arXiv:2302.00523]
Preprints
2024
[]		DynSUP: Dynamic Gaussian Splatting from An Unposed Image Pair (W Li, W Chen, S Qian, J Chen, D Cremers and H Li), In arXiv preprint arXiv:2412.00851, 2024.  [bibtex] [pdf]
Conference and Workshop Papers
2025
[]		AnyCam: Learning to Recover Camera Poses and Intrinsics from Casual Videos (F Wimbauer, W Chen, D Muhle, C Rupprecht and D Cremers), In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025. ([project page]) [bibtex]
2024
[]		Leveraging neural radiance fields for uncertainty-aware visual localization (L Chen, W Chen, R Wang and M Pollefeys), In IEEE International Conference on Robotics and Automation (ICRA), 2024.  [bibtex] [arXiv:2310.06984]
[]		LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry (W Chen, L Chen, R Wang and M Pollefeys), In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. ([project page]) [bibtex] [arXiv:2401.01887]