Max Robotics — Global Service Robot Directory + US Certification

About

Viktor Larsson is a leading researcher at the intersection of computer vision, privacy, and robotics, with a primary focus on privacy-preserving 3D reconstruction and visual localization. His most impactful work, "Privacy Preserving Structure-from-Motion" (2020, 36 citations), pioneers methods to perform 3D scene reconstruction without exposing sensitive image content, addressing critical privacy concerns in cloud-based mixed reality and robotics applications. Larsson further advanced this field with "Privacy Preserving Localization and Mapping from Uncalibrated Cameras" (2021, 9 citations), overcoming the fundamental limitation of requiring calibrated cameras—a key step toward practical, privacy-aware systems. His recent contributions include "Learning-Based Dimensionality Reduction for Computing Compact and Effective Local Feature Descriptors" (2023, 6 citations), which develops efficient, learned descriptors that maintain high performance in image matching, retrieval, and localization tasks. By combining privacy engineering with state-of-the-art feature learning, Larsson’s work enables secure, scalable solutions for real-world deployment in AR/VR and autonomous systems, making him a notable figure in privacy-preserving computer vision.