Developing a Distributed Computing Framework for Accelerating Deep Learning for Google Street View
Digital maps has been increasingly used in the past few years, and street views in maps potentially draw people's attention before they arrive their destination. Accordingly, study the human opinion to a street view of specific address is attractive as it is able to provide useful information for sociology and psychology in the near future. However, the original deep learning (DL) model used in processing street views can barely release all GPUs computational power, and lack of substantial scalability and stability when encountering relatively large dataset. To address these issues, we build a distributed computing framework to accelerate the DL model in server clusters which improves the GPU utilitization significantly. Along with the framework, an application is developed to fetch historical street views which will be fed into the DL model from Google APIs. A data analyzing component is also implemented to enhance data collection and summary. With the developed framework, the processing efficiency in server clusters of the DL model is noticeably improved which brings considerably shorter prediction time.
Major Advisor: Lizhong Chen
Committee: Perry Hystad
Committee: Yue Zhang
Friday, September 20, 2019 at 12:30pm to 2:30pm
Kelley Engineering Center, 2057
110 SW Park Terrace, Corvallis, OR 97331