Activity Recognition

Project ID
FG-567
Project Lead
Project Categories
Computer Science
Project Keywords
Project Members
Abstract
We are studying the connection of activity recognition from multiple viewpoints with various neural network architectures. Specifically, we are trying to learn a joint model to understand how activities are represented in different feature spaces. Michael Ryoo is my advisor for the project.
Use of FutureSystems
We intend to use the machine to train several activity recognition models.
Scale of Use
About 5000 GPU hours