Exploiting Redundancy for Reliable Aerial Computer Vision

This talk will highlight some recent results in deriving information (3D reconstruction, semantic segmentation, classification of objects) from a set of highly redundant images (acquired by air-planes or micro-aerial vehicles). The primary focus will be on the exploitation of the highly redundant data that is now available. We will demonstrate that redundancy is the key for fully automatic data processing, enabling a variety of large scale applications.