Kalle Åström

Professor in Mathematics, Centre for Mathematical Sciences, Lund University
Computer vision · machine learning · geometry · AI

I am a professor at the Centre for Mathematical Sciences at Lund University. My research spans mathematical foundations and practical methods for computer vision and machine learning, with a particular interest in the connections between geometry, learning and sensing.

I enjoy working across the full range from mathematical theory to algorithms, applications and collaboration with industry and other disciplines.

News

September 2026
I was General Chair for ECCV 2026 in Malmö. Great fun with 7000 researchers at Malmö Arena and Malmömässan.

2026
Recent work includes room layout estimation, synthetic cohorts for Alzheimer’s disease research, synthetic tau-PET generation and explainable deep learning for medical imaging.

Research

My current research interests include:

• Computer vision and geometric vision — multiple-view geometry, structure from motion, camera calibration, pose estimation and geometric reasoning.
• Machine learning and AI — combining data-driven learning with mathematical models, geometry and physical structure.
• Spatial sensing and localisation — mapping, localisation and positioning using vision, sound, radio and other sensors.
• Medical image analysis — machine learning and image analysis for applications including prostate cancer, Alzheimer’s disease and cardiovascular imaging.
• Mathematical modelling — algebraic, probabilistic and optimisation-based methods for perception, inference and decision-making.

A recurring theme is the interaction between mathematical structure and data-driven methods: understanding what can be solved from geometry and statistics, while using modern machine learning where it can improve robustness, scalability or performance.

Current projects

Some of the projects and research environments I am currently involved in include:

• Next Generation Spatial AI (ELLIIT)
• Synthetic high-dimensional cohorts for studies of neurodegenerative diseases (WASP/DDLS)
• Natural and Artificial Cognition at Lund University
• AI Lund — the Lund University network for artificial intelligence
• Incremental learning agents and human interaction (WASP Industrial PhD)
• Differentiable Neural Acoustics (WASP Academic PhD)
• Building a Machine to Sort DNA and Soft Particles (NanoLund + eSSENCE)
• Microfluidics with Machine Learning to Sort Cells and DNA (NanoLund)
• Machine learning methods for parsing symptoms and treatment responses in Parkinson’s disease (eSSENCE)

Group

I supervise a research group at the intersection of mathematics, computer vision, machine learning and sensing.

Current PhD students

Main supervisor:
Jens Gulin, Oskar Lindholm, Armon Haghighi

Co-supervisor:
Johanna Engman, Jennie Karlsson, Salma Rashed, M. Larsson, Gustav Hanning, Jamie Scotcher, Marcus Salzer Lucas Ferreira,

I have supervised PhD students working on topics ranging from geometric computer vision and polynomial solvers to audio localisation, sensor calibration, medical image analysis and machine learning.

Publications

I have published around 300 refereed scientific papers in computer vision, image analysis, machine learning, geometry, signal processing and related applications.

Selected recent publications

• PolyLayout: Multi-room Manhattan Layout Estimation — ECCV 2026.
• Benchmarking privacy and utility in synthetic tabular cohorts for Alzheimer’s disease research — 2026.
• Generating synthetic tau-PET scans in Alzheimer’s disease from MRI, blood biomarkers and demographics with deep learning — 2026.
• Sparse Multiview Open-Vocabulary 3D Detection — ICCVW 2025.
• Algebraic analysis of Doppler-based positioning with application to LEO satellites — IPIN 2025.
• Single-Source Localization as an Eigenvalue Problem — IEEE Transactions on Signal Processing, 2025.
• Detection and localization of drones and UAVs using sound and vision — CVPR 2025.
• NeuroNCAP: Photorealistic Closed-Loop Safety Testing for Autonomous Driving — ECCV 2024.

Google Scholar · Lund University Research Portal

Teaching

I teach and supervise at undergraduate, graduate and PhD level, particularly in mathematics, image processing and computer vision.

My PhD students are normally based in mathematics, and many of the research projects combine mathematical courses with work in computer vision, machine learning and applications.

Academic leadership and service

I currently coordinate the Natural and Artificial Cognition profile area at Lund University and the AI Lund network. I am also a board member of the Swedish AI Society (SAIS).

I have served as an editor, reviewer and programme committee member for international journals and conferences in computer vision and related areas.

From mathematics to applications

A large part of my work has involved collaboration beyond traditional mathematical research. This includes medical imaging, localisation and positioning, autonomous systems, audio and radio sensing, agriculture and animal behaviour.

I have also been involved in technology transfer and entrepreneurship, and have co-founded companies including Decuma, Cognimatics, Spiideo, Neuromathic and Inceptron.

Contact

Kalle Åström
Centre for Mathematical Sciences
Lund University
Sölvegatan 18
221 00 Lund, Sweden

karl.astrom@math.lth.se
Google Scholar · ORCID · Lund University profile