Roman Rädle
Also known as Roman Raedle
Roman Rädle is a researcher whose path runs from human-computer interaction to Meta's Segment Anything models for images and video. His 2017 dissertation at the University of Konstanz studied spatial and cross-device interactions and covers HuddleLamp, a low-cost sensing technology that tracks mobile devices on a table. In 2018, he left a postdoc at Aarhus University's Centre for Digital Creativity for an assistant professorship in its Computer Science Department. At Meta FAIR he was a core contributor to SAM 2 in 2024, which received an Honorable Mention among the Outstanding Paper Awards at ICLR 2025, and in November 2025 he co-authored SAM 3 under the affiliation Meta Superintelligence Labs.
- Badges
Timeline
-
2017
A dissertation on cross-device interaction
He defends his dissertation, Designing UbiComp Experiences for Spatial Navigation and Cross-Device Interactions, at the University of Konstanz in January 2017. The thesis covers HuddleLamp, a low-cost sensing technology that tracks multiple mobile devices on a table, and the Blended Shelf and Integrative Workplace prototypes for library knowledge work.
-
2018
Assistant professor at Aarhus
In 2018, he moves from a postdoc at Aarhus University's Centre for Digital Creativity, which ends in December 2018, to an assistant professorship in the university's Computer Science Department. His work there focuses on tools for education that work across a user's device ecology.
-
2024
SAM 2 at Meta FAIR
In 2024, he is one of 18 authors of SAM 2: Segment Anything in Images and Videos, with a Meta FAIR affiliation as a core contributor. The foundation model for promptable visual segmentation comes with a data engine that collects the largest video segmentation dataset to date. In image segmentation, it is more accurate and 6x faster than the original Segment Anything Model.
-
2025
ICLR 2025 honorable mention
SAM 2 receives an Honorable Mention among the Outstanding Paper Awards at ICLR 2025, one of 3 honorable mentions chosen from an initial pool of 36 papers.
-
November 2025
SAM 3 and concept prompts
In November 2025, he is among the 38 authors of SAM 3: Segment Anything with Concepts, a unified model that detects, segments and tracks objects in images and videos from concept prompts such as short noun phrases or image exemplars. His affiliation is Meta Superintelligence Labs.
Career
- From 2025Meta Superintelligence Labs
- From 2024FAIR (Meta)Core contributor
- 2018Aarhus UniversityPostdoctoral researcher, Assistant Professor
Education
- 2017University of KonstanzPhD
Sources and credits
Photo: We have no published photo of Roman Rädle yet. Send us a free-licensed photo.
- Designing UbiComp Experiences for Spatial Navigation and Cross-Device Interactions (KOPS, Universität Konstanz)
- Roman Rädle is leaving the CDC
- SAM 2: Segment Anything in Images and Videos (arXiv abstract)
- SAM 2: Segment Anything in Images and Videos (arXiv HTML)
- Announcing the Outstanding Paper Awards at ICLR 2025
- SAM 3: Segment Anything with Concepts (arXiv abstract)
- SAM 3: Segment Anything with Concepts (arXiv HTML)



