Home > Servers > PowerEdge Components > White Papers > Developing and Deploying Vision AI with Dell and NVIDIA Metropolis > Overview
The NVIDIA Metropolis framework accelerates the development and deployment at scale of AI-enabled video-analytics applications. It consists of three entities:
The following figure shows the use of this end-to-end pipeline. During the deployment phase, TensorRT CUDA libraries are used, along with the multi-instance and multi-GPU capabilities of the Triton Inference Server. This usage provides inference at scale on GPU-based infrastructure (cloud, on-premises data center, or edge). Figure 2 shows this AI vision ecosystem, including Dell Technologies as an NVIDIA EGX Server Partner.