Conferences and supporting programme
Streaming Analytics Framework (SAF)
Abstract: With a prediction of over 13 billion cameras connected to the Internet by 2030, the abundance of visual data opens vast analytics opportunities, but with a caveat of that the outstanding data volume prevents from fully utilizing the well-developed cloud computing infrastructure whereas the demanding computation is a showstopper kills edge only solutions. Analytics aware offloading between edge and cloud is needed. Streaming analytics framework (SAF) is an open source framework for edge-to-cloud streaming analytics that optimizes for low-latency continuous inference workloads, co-developed by Intel and Intel Science and Technology Center for Visual Cloud Systems (ISTC-VCS). In this paper, we will present the current status and future plan of SAF, with its key differentiated features, including (1) high-level APIs for creating workloads by general directed graph composition, (2) seamless Intel hardware optimization using OpenVINO among various Intel toolchains, (3) transparent scaling by automated edge-to-cloud workload orchestration.
--- Date: 27.02.2019 Time: 4:00 PM - 4:30 PM Location: Conference Counter NCC Ost