Deep Learning @ UGA

Learning On-Air Hand Gestures from Wi-Fi Signals on Smartphones

We introduce a new online machine learning solution called “Wisture” for recognizing on-air hand gestures on a smartphone. Using the standard Wi-Fi signal measurements, we apply the Long Short-Term Memory (LSTM) based learning as our inference mechanism to classify the different types of gestures made over the smartphone. Wisture solution is unique in a way that it works out-of-box in terms of both hardware and software. In this talk, we present the characteristics of Wisture along with its experimental evaluation in different scenarios demonstrating high accuracy in detecting and classifying three gestures.

Ramviyas Parasuraman is an Assistant Professor in the Department of Computer Science at the University of Georgia, Athens, USA. His research mainly focuses on heterogeneous multi-robot systems, networked autonomous vehicles, intelligent robot teleoperation, and human-robot interfaces. Previously, he has held postdoctoral researcher positions at Purdue University and KTH Royal Institute of Technology. He conducted his doctoral research at the European Organization for Nuclear Research (CERN) and obtained his Ph.D. from Universidad Politécnica de Madrid. He was a recipient of the prestigious Marie Skłodowska Curie ESR fellowship and is an alumni of the Indian Institute of Technology Delhi.

Presentation Slides