
This video program is a part of the Premium packages:
High-Bandwidth Low-Latency Applications using AI and Edge Computing Infrastructure – A Snapshot
- IEEE MemberUS $1.00
- Society MemberUS $0.00
- IEEE Student MemberUS $1.00
- Non-IEEE MemberUS $2.00
- IEEE MemberUS $50.00
- Society MemberUS $0.00
- IEEE Student MemberUS $25.00
- Non-IEEE MemberUS $100.00
High-Bandwidth Low-Latency Applications using AI and Edge Computing Infrastructure – A Snapshot
This talk reviews experiences in pursuing experimental studies with high-bandwidth low-latency applications which utilize video-based deep learning. Using the example of the COSMOS smart-city intersection, we illustrate one approach to privacy preservation, and show how it affects the results in object detection accuracies. We discuss the development/tool chain, debugging, profiling and evaluation of the results, and provide some insight into what challenges will have to be further overcome to facilitate support for reliable low-latency high-bandwidth applications.
High-Bandwidth Low-Latency Applications using AI and Edge Computing Infrastructure – A Snapshot
Zoran Kostic, Columbia University, NYC