Toward large-scale situation awareness applications on camera networks

Ubiquitous deployment of cameras and recent advances in video analytics enable a new class of applications, situation awareness using camera networks. The application class includes surveillance, traffic monitoring, and assisted living that autonomously generate actionable knowledge from a large number of camera streams. Despite technological advances, developing a large-scale situation awareness application still remains a challenge due to the programming complexity, highly dynamic workloads, and latency-sensitive quality of service. To solve the problem, my research topic concerns developing a programming model and a runtime system to support large-scale situation awareness applications on camera networks. The programming model requires a minimal set of domain-specific handlers from the domain experts, allowing them to focus on the algorithmic aspect of situation awareness applications rather than the details of distributed programming. The runtime system provides automatic resource management using smart cameras and the cloud for handling dynamic workloads and ensuring latency-sensitive quality of services.