Abstract | ||
---|---|---|
We present a novel action recognition system that is able to learn how to recognize and classify actions. Our system employs a three-layered neural network hierarchy consisting of two self-organizing maps together with a supervised neural network for labelling the actions. The system is equipped with a module that preprocesses the 3D input data before the first layer, and a module that transforms the activity elicited over time in the first layer SOM into an ordered vector representation before the second layer, thus achieving a time invariant representation. We have evaluated our system in an experiment consisting of ten different actions selected from a publicly available data set with encouraging result. |
Year | DOI | Venue |
---|---|---|
2017 | 10.5220/0006199305830590 | ICAART: PROCEEDINGS OF THE 9TH INTERNATIONAL CONFERENCE ON AGENTS AND ARTIFICIAL INTELLIGENCE, VOL 2 |
Keywords | Field | DocType |
Self-organizing Maps,Neural Networks,Action Perception,Hierarchical Models | Data mining,Computer science,Action recognition,Self-organizing map,Artificial intelligence,Artificial neural network,Hierarchy,Trajectory,Machine learning,Robotics | Conference |
Citations | PageRank | References |
1 | 0.36 | 0 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Zahra Gharaee | 1 | 4 | 2.80 |
Peter Gärdenfors | 2 | 1699 | 183.78 |
Magnus Johnsson | 3 | 99 | 13.51 |