Using deep learning to predict human movement
What if we could use technology to simulate and predict human movement, and in turn, use that information to improve evacuation, prevent accidents and ease traffic congestion in times of crisis?
Associate Professor Gary Tan from NUS Computing has been doing just that. He has developed a special framework that uses deep learning methods to track real-life movement of pedestrians through video feeds. This behaviour is then translated into data for a virtual simulator to reconstruct scenes and events that are too costly or dangerous to replicate.
Using this data-driven approach offers a more accurate prediction of human responses during a crisis situation, and facilitates the design of more optimised strategies for crowd control.
Read more here.