Research Areas
Event-based Vision
Our lab specializes in developing cutting-edge algorithms for event-based cameras, which represent a paradigm shift in computer vision. We focus on continuous video reconstruction, motion understanding, and high-speed vision applications using event cameras.
- Continuous color video reconstruction
- Motion field estimation
- Low-latency visual processing
- Neural event processing
3D Computer Vision
We develop novel algorithms for 3D reconstruction using event cameras. Our research focuses on converting event-based apparent contours into accurate 3D models, enabling real-time reconstruction of dynamic scenes.
- Event-based 3D reconstruction
- Continuous visual hull computation
- Dynamic scene understanding
- Real-time 3D modeling
Motion Understanding
Our research focuses on understanding complex motion patterns in dynamic scenes using event cameras. We develop unsupervised learning approaches for motion segmentation and scene understanding, particularly useful for robotics applications.
- Independent motion segmentation
- Unsupervised learning
- Dynamic scene analysis
- Event-based tracking
Current Projects
Neural Event Processing
Developing neural network architectures specifically designed for processing event-based data streams.
Dynamic Scene Understanding
Creating algorithms for understanding complex, dynamic scenes using multi-modal sensor fusion.
Embodied Navigation
Investigating active perception strategies for robot navigation in unknown environments.