Cracking open the DNN black-box: Video Analytics with DNNs across the Camera-Cloud Boundary

Workshop on Hot Topics in Video Analytics and Intelligent Edges (HotEdgeVideo) |

Télécharger BibTex

Advancements in deep neural networks (DNNs) and widespread deployment of video cameras have fueled the need for video analytics systems. Despite rapid advances in system design, existing systems treat DNNs largely as “black boxes” and either deploy models entirely on a camera or compress videos for analysis in the cloud. Both these approaches affect the accuracy and total cost of deployment. In this position paper, we propose a research agenda that involves opening up the black box of neural networks and describe new application scenarios that include joint inference between the cameras and the cloud, and continuous online learning for large deployments of cameras. We present promising results from preliminary work in efficiently encoding the intermediate activations sent between layers of a neural network and describe opportunities for further research.

Microsoft Rocket: Hybrid Edge + Cloud Video Analytics Platform

Today, video cameras are being used at a large scale by public and private enterprises for a variety of reasons—from security surveillance and traffic planning to consumer support in retail and hospitality settings. Thanks to gains in computer vision, particularly object detection and classification, video analysis has become more accurate. Fast and affordable real-time analysis, however, is lagging. Project Rocket seeks to make easy, cost-effective video analysis of live camera streams a reality. Project Rocket, an extensible software stack that leverages the edge and cloud, is designed with maximum functionality in mind, capable of meeting the needs of varying video analytic applications. In this webinar, Microsoft researchers Ganesh Ananthanarayanan and Yuanchao Shu explain how Rocket—now open source on GitHub—uses approximation to run scalable analytics across…