Integrating recognition and reconstruction for cognitive scene interpretation
B. Leibe

 

In this talk, we will present recent work on building a cognitive feedback loop which integrates real-time geometry estimation with appearance-based object detection. We demonstrate how those two components can benefit from each other's continuous input and how the transferred knowledge can be used to improve scene analysis. Apart from improving both recognition performance and overall system robustness, the proposed tight integration enables the construction of novel capabilities such as accurate 3D estimation of object locations and their temporal integration in a world coordinate frame.

 

presentation

videos:

city1.mpg
vid8.mpg
video-3dhypos-final.mpg
video-3dhypsos-online.mpg
video-carving.mpg
video-detections-gp.mpg
video-integration.mpg
video-textured-reconstructionl.mpg