Year: 2020
Type: dissertation
Abstract: A key determinant of a video analytics application success lies in its ability to operate in a real-time fashion and must do so, regardless of the number of video streams the system needs to process. Hence, there is a need for a scalable and distributed video analytics system that can leverage the power of the cloud and thus successfully deliver on its promise. This work addresses this challenge by taking a bottom-up approach in defining the required system architecture and design for enabling large-scale execution of video analytics jobs in the cloud in a distributed and scalable fashion. The proposed system is designed by leveraging the existing open-source big data technologies and is deployed and evaluated to shed light on several intrinsic aspects of modern real-time video analytics applications.
Source: TSpace (University of Toronto)
Author: Seyed Mohammad Pakdaman
Institution: University of Toronto
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: University of Toronto
OpenAlex ID: W3156495308
