Build and deploy AI models using NVIDIA DeepStream on Jetson and AWS IoT Core

Edge computing is essential for processing and generating real-time insights from billions of sensors around the world. Pixels from cameras are converted to actionable insights using AI on the edge.

Intelligent video analytics technology is used to aid traffic management for cities, automated checkout in retail stores, non-contact automated visual inspection in manufacturing facilities, or crowd monitoring for safe social distancing measures and more. Once the insights are generated on the edge, the next step is to send this vital information to the cloud for further analytics. 

NVIDIA and Amazon Web Services (AWS) have partnered to simplify workflows to  quickly build, deploy, and manage AI-based video analytics applications. With NVIDIA NGC, data scientists and developers can deploy AI frameworks with containers, get a head start with pre-trained models, and deploy AI services in cloud native containers and orchestrate using Kubernetes which can be effectively deployed at the edge and to the cloud giving you faster time-to-market.

NVIDIA DeepStream is the accelerated framework to build AI-powered Intelligent Video Analytics apps and services. With DeepStream end-to-end inference and IoT capabilities, developers can generate real-time insights on the edge and seamlessly connect to the cloud. The SDK offers a great degree of flexibility to the developers; enabling them to customize their video analytic solution. 

To connect to the cloud, developers can use the message-broker plugin to either use one of the built-in message protocols or create a custom adapter.  

DeepStream Graph Architecture

The message broker plugin provides the API to create and integrate custom protocol adapters to communicate to any cloud services. AWS created a custom adapter to publish MQTT messages from DeepStream applications running on the edge to AWS IoT Core. This enables users to deploy and manage AI applications on the edge using AWS cloud services.    

Solution Overview 

For a detailed solution and workflow walkthrough, we’ve created a joint webinar and tutorial on “Build with DeepStream, deploy and manage with AWS IoT services “ which is now available on-demand.