Martin R. Oswald
Martin R. Oswald is a computer vision researcher whose work concerns three-dimensional scenes. He studied computer science in Dresden and Sydney, completed an engineering master’s degree in Chile and earned a computer vision doctorate at TU Munich in 2015. He researched at Amsterdam and ETH Zurich. He coauthored Point-SLAM, a neural method that uses a growing point cloud to represent a scene while estimating camera motion. The method was evaluated on established research datasets for tracking, mapping and rendering.
- Now at
- ETH Zurich and University of Amsterdam
- Works
- 129
- Citations
- 3,464
- h-index
- 29
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2007–2008
Two engineering paths
Oswald obtained a computer science Diplom from TU Dresden in 2007 after study in Dresden and Sydney. In 2008 he completed a civil engineering master’s degree at Federico Santa Maria in Chile with DAAD support.
-
2015
Doctoral work in computer vision
He earned his computer vision doctorate at TU Munich in 2015 under Daniel Cremers. His research field is three-dimensional computer vision, and he has held roles at Amsterdam and ETH Zurich.
-
2023
A neural point cloud for mapping
Oswald coauthored Point-SLAM with Erik Sandström, Yue Li and Luc Van Gool. The system builds a point cloud from color and depth images and uses a shared neural scene representation for camera tracking and mapping.
-
2023
Adapt the scene representation
Point-SLAM changes point density according to the detail in the input. The authors report tracking, mapping and rendering experiments on Replica, TUM-RGBD and ScanNet.
Sources and credits
Photo: ellis.eu; Martin R. Oswald profile photograph; creator not stated. License: Not stated by source. Source: ellis.eu. Cropped to a square and resized. Ask us to remove or replace this photo.