Fredrik Kahl
Fredrik Kahl is a computer vision researcher whose work centers on geometric deep learning and 3D scene understanding. He earned a mathematics PhD at Lund University, won the 2001–2002 Best Nordic Thesis Award, and served as a professor at Lund from 2006 to 2018. In 2005, he won the ICCV Marr Prize for “Globally Optimal Estimates for Geometric Reconstruction Problems.” His 2025 spherical reconstruction paper developed a global method for uncalibrated camera motion, and a 2026 paper applied equivariance to make Vision Transformers more efficient. As of October 2026, he is a professor and unit head at Chalmers and a 2026–2028 Hi!Paris Visiting Chair at École des Ponts ParisTech.
Timeline
-
1997–2001
A mathematics PhD
He earns a PhD in mathematics at Lund University from 1997 to 2001.
-
2001–2002
A Nordic thesis award
His thesis, Geometry and Critical Configurations of Multiple Views, wins the Best Nordic Thesis Award in 2001–2002.
-
2005–2007
The ICCV Marr Prize
With Didier Henrion, he wins the ICCV Marr Prize in 2005 for “Globally Optimal Estimates for Geometric Reconstruction Problems.” They publish a paper with that title in the International Journal of Computer Vision in 2007.
-
2006–2018
Professor at Lund
His career includes postdoctoral work at UC San Diego and the Australian National University. He serves as a professor at Lund University from 2006 to 2018.
-
2025
Spherical structure from motion
With Jonathan Ventura and Viktor Larsson, he coauthors “Uncalibrated Structure from Motion on a Sphere” for ICCV 2025. Their method builds global structure-from-motion for uncalibrated spherical camera motion and shows self-calibration from three or more views.
-
2026
Equivariant vision transformers
With David Nordström and Johan Edstedt, he coauthors Quick ViTs in 2026, an equivariant vision-transformer method. Its octic equivariant linear layers reduce FLOPs by 5.33 times and memory by up to 8 times compared with ordinary linear layers. His research focuses on geometric deep learning and 3D scene understanding.
-
As of October 2026
Chalmers professor and visiting chair
He is a full professor and Head of Unit in Signal Processing and Biomedical Engineering at Chalmers, where he leads the Computer Vision Group. From 2026 to 2028, he serves as a Hi!Paris Visiting Chair at the IMAGINE research group, École des Ponts ParisTech.
Sources and credits
Photo: We have no published photo of Fredrik Kahl yet. Send us a free-licensed photo.
- Fredrik Kahl – Computer Vision and Machine Learning
- Homepage of Changyin Zhou
- Globally Optimal Estimates for Geometric Reconstruction Problems - Lund University
- Uncalibrated Structure from Motion on a Sphere
- Quick ViTs: Speeding up Vision Transformers Through Equivariance
- Fredrik Kahl | Chalmers
- 11 International Visiting Chairs Join Hi! PARIS in 2026