Object Discovery
 |
Object discovery is the task of finding unknown objects in a scene. Humans can easily distinguish objects from background without apparent effort, however, it is a challenging problem in machine vision with a chicken and egg problematic: how to tell what the objects are if their properties and features are not known? Our approaches base on concepts of the human visual system such as saliency detection, segmentation, and figure-ground segregation to generate object candidates. This biological inspiration is combined with computer vision methods to generate new algorithms such as finding partial maximum spanning trees in graphs that represent the image (ICVS 2015). The edge weights in these graphs are computed by saliency and similarity measures. Our object discovery methods are especially well suited for complex real-world scenes that contain numerous objects as the data acquired from mobile vision systems, e.g. robots. |
 | Graph-based Object Discovery in RGB-D Data
Saliency-Guided Object Candidates Based on Gestalt Principles |
|---|---|
 | Object discovery & tracking in videos (ICRA 2015,
Sequence-level Object Candidates Based on Saliency for Generic Object Recognition on Mobile Systems |
 | Object discovery in RGB-D frames (ICRA 2015):Saliency-based Object Discovery on RGB-D Data with a Late-Fusion ApproachGermán Martín García, Ekaterina Potapova, Thomas Werner, Michael Zillich, Markus Vincze, and Simone Frintrop Datasets and results |
 | Object discovery in images (ICPR 2014):A Cognitive Approach for Object DiscoverySimone Frintrop, Germán Martín García, and Armin B. Cremers |
 | Object discovery and scene exploration in 3D (CogSci 2013):A Computational Framework for Attentional 3D Object DetectionGermán Martín García and Simone Frintrop |