| ECVision indexed and annotated bibliography of cognitive computer vision publications This bibliography was created by Hilary Buxton and Benoit Gaillard, University of Sussex, as part of ECVision Specific Action 8-1 The complete text version of this BibTeX file is available here: ECVision_bibliography.bib
Loos, H.S. and von der Malsburg, C.
ABSTRACT
We present a method which continuously learns representations of arbitrary objects. These object representations can be stored with minimal user interaction (1 Click Learning). Appropriate training material has the form of image sequences containing the object of interest moving against a cluttered static background. Using basically the method of unsupervised growing neural gas modified to adapt to non stationary distributions on binarized difference images, a model of the moving object is learned in real time. Using the learned object representation the system can recognize the object or objects of the same class in single still images of different scenes. The new samples can be added to the learned object model to further improve it. |
|
| Site generated on Friday, 06 January 2006
|