Classifying Patterns of Visual Motion - A Neuromorphic Approach
Abstract
We report a system that classifies and can learn to classify patterns of visual motion on-line. The complete system is described by the dynam- ics of its physical network architectures. The combination of the fol- lowing properties makes the system novel: Firstly, the front-end of the system consists of an aVLSI optical flow chip that collectively computes 2-D global visual motion in real-time [1]. Secondly, the complexity of the classification task is significantly reduced by mapping the continu- ous motion trajectories to sequences of ’motion events’. And thirdly, all the network structures are simple and with the exception of the optical flow chip based on a Winner-Take-All (WTA) architecture. We demon- strate the application of the proposed generic system for a contactless man-machine interface that allows to write letters by visual motion. Re- garding the low complexity of the system, its robustness and the already existing front-end, a complete aVLSI system-on-chip implementation is realistic, allowing various applications in mobile electronic devices.
Cite
Text
Heinzle and Stocker. "Classifying Patterns of Visual Motion - A Neuromorphic Approach." Neural Information Processing Systems, 2002.Markdown
[Heinzle and Stocker. "Classifying Patterns of Visual Motion - A Neuromorphic Approach." Neural Information Processing Systems, 2002.](https://mlanthology.org/neurips/2002/heinzle2002neurips-classifying/)BibTeX
@inproceedings{heinzle2002neurips-classifying,
title = {{Classifying Patterns of Visual Motion - A Neuromorphic Approach}},
author = {Heinzle, Jakob and Stocker, Alan},
booktitle = {Neural Information Processing Systems},
year = {2002},
pages = {1147-1154},
url = {https://mlanthology.org/neurips/2002/heinzle2002neurips-classifying/}
}