TITLE: Video Data Management for the Real World
The proliferation of cameras deployed throughout our world is enabling and accelerating exciting new applications, such as virtual and augmented reality, autonomous driving, drone analytics, and smart infrastructure. However, because cameras produce large volumes of content-rich data, application development remains tedious. In this talk, I describe several video data management systems designed to simplify application development and optimize execution. By exposing declarative interfaces, these systems automatically produce efficient execution strategies that include leveraging heterogeneous hardware, operating directly on the compressed representation of video data, and improving video storage performance. Through these systems, we show that the application of fundamental data management principles to this space vastly improves performance while greatly decreasing application development complexity.
Brandon Haynes is a Ph.D. candidate in the Paul G. Allen School of Computer Science at the University of Washington, where he is advised by Magdalena Balazinska and Alvin Cheung. His research interests focus on designing and building data management systems that address the needs of new and emerging application domains. His recent work has focused on video analytics and video data management, focusing especially on areas with extremely large data sizes (e.g., VR/AR and large-scale multi-camera networks) and new application areas (e.g., deep learning). His research has been supported in part by a University of Washington fellowship and the UW Reality Lab.