Holistic Object Recognition using Neural Networks

In this project we are developing a system that is aimed to overcome a major limitation of current computer vision: The specialization of vision architectures to one special task. In this approach, artificial neural networks (ANN) are applied to learn the appearance of objects from samples images. By this means, the costly designing of geometric object models can be avoided. This work is realized in the framework NESSY (NEural viSion SYstem), a software package that allows an easy design and visualization of image processing systems. We apply a neural architecture for a combined, trainable feature extraction and knowledge representation, together with modules for focus of attention that allow(s) the decomposition of even complicated scenes. We are developing a system for automated image data acquisition and unsupervised exploration of image data structures to facilitate a semi-automatic categorization of large image databases. This system could be applied to several [different image domains] and is part of the [hybrid recognition system] of [SFB 360]. A public demonstration of NESSY was given in the Heinz Nixdorf MuseumsForum, the largest computer museum of the world.