SIGCOMM 2020 - Server-Driven Video Streaming for Deep Learning Inference

Explore server-driven video streaming techniques for deep learning inference in this conference talk from SIGCOMM 2020. Delve into the intricacies of video streaming for analytics, real-time camera set heuristics, and server-driven approaches. Witness a demonstration of object detection and examine experimental results that showcase the effectiveness of these methods. Gain insights into the potential overhead and participate in a Q&A session to further understand the implications of this innovative approach to video streaming for deep learning applications.

Introduction
Video Streaming for Analytics
RealTime Camera Set Heuristics
ServerDriven Video Streaming
Demo
Object Detection
Experimental Results
Conclusion
Questions
Overhead
QA