- Field
- Computer vision
- Task
- Instance segmentation
- Data
- Satellite imagery, sparse labels
- Approach
- DINOv3 self-supervised backbone
Fellowship
The Fellowship Program puts a team of technical students from Lund University on one real problem for a nonprofit, for a semester. The work is open source, the nonprofit keeps what the team builds, and Lund AI Society scopes the problem and runs the project.
Spring 2027 · 10 seats
Current projects
- Field
- Natural language processing
- Task
- Information extraction
- Data
- Multilingual case notes, few labels
- Approach
- Synthetic data, on-premise
Spring 2027
1 Dec
Applications close.
25 Jan
Kickoff.
28 May
Projects end.