Sebastian Jung

I'm a researcher and PhD student at the German Aerospace Center (DLR) in Munich under the supervision of Rudolph Triebel. I am interested in object-centric computer vision, geometric understanding and few-shot/zero-shot learning. My focus is motivated by its applications in robotics and augmented reality.

Email / Scholar / LinkedIn / X / Github

News

03/26: Our TAPNext++ paper was accepted to CVPR Findings 2026 — written during my internship at Google
11/25: Our NeMO paper was accepted to 3DV 2026 as an Oral Presentation
08/25: Started working at Google as a Student Researcher in the group of Federico Tombari under supervision of Martin Sundermeyer
04/25: Started my PhD at KIT under Rudolph Triebel
01/23: Started working at the German Aerospace Center as a Computer Vision and Robotics Researcher

Publications

TAPNext++: What's Next for Tracking Any Point?
Sebastian Jung, Artem Zholus, Martin Sundermeyer, Carl Doersch, Ross Goroshin, David Joseph Tan, Sarath Chandar, Rudolph Triebel, Federico Tombari
CVPR Findings, 2026
Project Page / arXiv / Code

State-of-the-art point tracking model for long video sequences. Enables re-detection after occlusion via roll augmentation and distributed sequence-parallel training on 1024-frame videos, achieving up to 562 FPS throughput.

Finding NeMO: A Geometry-Aware Representation of Template Views for Few-Shot Perception
Sebastian Jung, Leonard Klüpfel, Rudolph Triebel, Maximilian Durner
3DV, 2026 (oral)
Project Page / arXiv / Code

An encoder-decoder architecture to generate a geometric, object centric representation. The decoder allows object detection, segmentation and pose estimation on any object without fine-tuning.