Your Browser is Out of Date

Nytro.ai uses technology that works best in other browsers.
For a full experience use one of the browsers below
Dell EMC Ready Solutions for AI: Safer and Smarter Environments with Intelligent Video Analysis
Executive summary Creating safer and smarter environments Solution overview
Introduction
Dell EMC PowerEdge servers
NVIDIA components
Deep Vision AI components
Camera details
Configuration Deployment and management Solution outcome Conclusion

Introduction

Introduction

  • This solution demonstrates how a Dell Technologies-designed infrastructure can support the capabilities of Deep Vision AI software to create safer and smarter retail, manufacturing, health care, and public services environments. We also describe Deep Vision's AI-based Software Modules. These modules receive input from new or existing cameras that are connected to the Dell Technologies supporting platform to support various analytics that are focused on improving the efficiency of business operations and keeping commercial and public environments safer.

    This Ready Solution consists of a single PowerEdge server with proximity to the camera deployed at the edge. It consists of the hardware and software components shown in the following figure:

     

    Figure 1.              Deep Vision AI Platform on NVIDIA EGX Software Stack and Dell PowerEdge Server

    The Ready Solution includes:

    • NVIDIA components—This solution uses the following NVIDIA components:
    • EGX software stack for AI workloads running at the edge. It includes Kubernetes, Helm, Tiller, and the NVIDIA GPU Operator for deploying and managing the GPU components necessary for GPU-enabled Kubernetes.
    • NVIDIA Metropolis, which is a collection of tools and technologies including DeepStream SDK. The toolkit includes the TensorRT SDK for high-performance deep learning inference. AI applications, like Deep Vision AI, can use DeepStream SDK and TensorRT to process video streams and optimize deep learning inference.
    • T4 GPU for video decoding and inference.
    • CUDA driver and toolkit for interfacing with the GPU.
    • Deep Vision AI Platform—The AI models from Deep Vision AI, which is deployed as a Docker container. This container uses the NVIDIA DeepStream SDK and the NVIDIA CUDA toolkit.

    You can also install Kubernetes container management software in a virtual machine (VM). VMware ESXi server can be installed on the PowerEdge server, and the Deep Vision AI platform is deployed as a container in that VM. You can install other applications as VMs to increase the server resource utilization. Deployment guidance for a virtualized environment is beyond the scope of this document.