Can I read Graph Convolutional Networks for Action Recognition on EtoBox?

Graph Convolutional Networks for Action Recognition by Jonas Robinson is a document available to read on EtoBox.

What is Graph Convolutional Networks for Action Recognition about?

The document proposes using neural architecture search to learn graph convolutional networks for human action recognition from skeleton data. It aims to address limitations of existing GCN methods that ignore implicit joint correlations and use fixed graph structures. The method searches a GCN space built with multiple graph function modules using a strategy for efficient sampling and memory usage. It is evaluated on standard human action recognition datasets, outperforming previous state-of-the-art methods

Author

Jonas Robinson

Language

EN