Visual Intelligence Beyond Closed Benchmarks
Video, actions, objects, scenes, and interaction under changing categories, contexts, and goals.
Collaborate
Research Collaboration
I am broadly open to collaborations, student projects, and thesis supervision across computer vision and AI. I am particularly interested in ambitious questions around visual intelligence, multimodal reasoning, embodied perception, and evaluation beyond closed benchmarks, while also welcoming ideas outside my current publication areas.
Directions
Broad Research Interests
Multimodal and Embodied AI
Links between vision, language, spatial reasoning, agents, robotics, physical environments, and human activity.
Reliable Methods and Meaningful Evaluation
Robustness, uncertainty, imperfect supervision, generalization, datasets, and evaluation protocols that reveal what models really understand.
Fit
Good Starting Points
Research
New Research Ideas
A paper idea, methodological question, benchmark gap, dataset opportunity, or application domain where visual perception and AI could matter.
Supervision
Student Projects and Theses
Bachelor, master, or visiting student topics are welcome across computer vision and AI; the strongest topics usually connect a clear research question with a feasible prototype or evaluation.
Message