Graduate Research Colloquium: Optimizing Real-Time Video Analysis for the Internet of Things
|Author:||Brian Gravelle University of Oregon|
|Date:||April 27, 2017|
In recent years computing has begun to develop a new and exciting paradigm: the Internet of Things (IoT), in which numerous everyday devices become connected to the internet to collect, process, and act on information. These devices will allow computing to merge seamlessly with the infrastructure and activities that make up our daily lives to improve quality of life for all people.
Before this ideal can be achieve, we must overcome numerous technical, economic, and societal challenges. On the technical side many of the devices incorporated into the physical world will be small computers with limited computation and energy resources. To address this limitation scientists and engineers will need to work to make existing algorithms and systems more efficient in terms of computation and energy use.
Early adopters of the IoT paradigm include city and traffic planners who use connected cameras to collect data and control the flow of traffic on various roadways. Such systems include traffic light control, tolling, and data collection. Our research focuses on the data collection aspect, exploring various methods of counting pedestrians, bicyclists, and cars. The research aims to compare and optimize algorithms in terms of computation time, accuracy, and power since different systems will have different needs in these three areas. As this is ongoing work methodology, but no results will be presented.