NVIDIA Metropolis Microservices

Ready to accelerate your vision AI journey from the edge to any cloud? This suite of powerful cloud-native microservices and reference applications will help you fast-track the development and deployment of your vision AI applications.

Get Started for Jetson Orin Get Started for Enterprise GPU

Metropolis microservices and reference applications diagram

Metropolis microservices give you powerful, customizable, cloud-native building blocks for developing vision AI applications and solutions—built to run on NVIDIA Cloud and data center GPUs, along with the NVIDIA Jetson Orin™ edge AI platform. These services let you unlock business insights for a wide range of spaces, ranging from retail stores and warehouses to airports and roadways. And they’re brought to life with reference applications that address a variety of challenging vision AI use-cases, including multi-camera tracking, occupancy heatmaps, AI-powered NVRs, and more.

Metropolis microservices make it far easier and faster to prototype, build, test, and scale deployments from edge to cloud with enhanced resilience and security.

Transform Your Workflow

Faster Time to Solution

Powerful, turnkey, API-driven, customizable vision AI building blocks accelerate application development.

Flexible Deployments

Cloud-native, industry-standard technologies enable seamless and scalable deployments anywhere, from NVIDIA Jetson Orin to any cloud.

Reusability Across Use Cases

Develop once and use across many applications with our modular microservice architecture.

Take Your Vision AI to the Next Level

Discover Real-Time Cross-Camera Insights

Reimagine your space with state-of-the-art, multi-camera tracking (MTMC) or vision-based real-time locating system (RTLS) technology. They go beyond single-camera perception to enable real-time insights for vast areas.

The NVIDIA system takes advantage of digital twins to simplify complex workflows, from synthetic data generation to model validation and facilitating rapid deployment. Discover patterns, optimize workflows, and enhance safety in your facilities with cutting-edge vision AI.

Read the blog

Metropolis fast tracks generative AI use cases

Productize Generative AI

Generative AI brings a new class of AI that enables models to understand the world in a more open way than previous methods. These models can comprehend natural language input and can give you a richer understanding of the scene. They’re also capable of delivering zero-shot and few-shot learning inference on images and videos.

The generative AI model can be seamlessly integrated with all the microservices to create a complete, deployable application. Metropolis Microservices make it easy to connect your generative AI application to live video stream and dynamically send natural language prompts remotely through REST API Endpoints.

Build Complete Vision AI Solutions

Metropolis microservices deliver a growing suite of options that include video storage and management, AI Inference pipelines, and analytics that go beyond perception and computer vision. These capabilities help describe what’s happening, reveal patterns, and predict what might happen next—all from objects’ movements and behaviors.

Metropolis microservices for Jetson also include platform services such as system monitoring, API gateway, cloud-connectivity, and more, providing faster time to production.

Supercharge your solution with a collection of microservices and reference applications. Enable sophisticated applications such as multi-camera tracking, AI-based NVR, zero-shot learning with generative AI, and more.

Metropolis microservices help you integrate vision AI app seamlessly with standard interfaces

Tap Into a Cloud-Native Microservice Architecture

This architecture enables the high flexibility, scalability, and maintainability that are increasingly important for the fast-moving world of AI solution development and deployment.

All the microservices and reference applications are built with industry standards like Docker, Kubernetes, Spark, Kafka, Redis, ELK stack, and others. In addition, they all have well-defined REST APIs to allow seamless integration with or within your existing applications and solutions.

Get Started With Metropolis AI Workflows

Discover cloud-native, pre-packaged reference applications to accelerate your AI solution development.

Multi-Camera Tracking

Track objects across large spaces with many cameras to improve operational efficiency and safety in places like retail stores, warehouses, and more.

Watch GTC Talk

Occupancy Analytics

Provide analytics on people movements such as space occupancy, time inside or outside ROI, movement patterns, etc. It also supports retail store analytics.

Few-Shot Product Recognition

Provide visual recognition capabilities that can adapt continuously to changing environments without needing much data.

Watch GTC Talk

AI NVR (Network Video Recorder)

Create an end-to-end insights and analytics application using services like data ingestion, video storage, inference, tracking, and more.

Learn More About AI NVR

Zero-Shot Detection Using Generative AI

Integrate generative AI capabilities, such as zero-shot learning, to transform your next-generation vision AI applications.

Read the Blog

Develop with Metropolis Microservices

Take advantage of these powerful and customizable cloud-native building blocks for your next vision AI solution.

SIgn Up for Developer Preview: Jetson | Enterprise GPU

		NVIDIA Jetson Orin	NVIDIA Enterprise GPU
Microservices	Video Storage Toolkit	✅	✅
	Scene AI Perception Service (NVIDIA DeepStream)	✅	✅
	Behavior Analytics	✅	✅
	Behavior Learning		✅
	Multi-Camera Tracking		✅
	Redis	✅
	Kafka		✅
	API Gateway	✅	✅
	Monitoring	✅	✅
	IoT Gateway	✅	✅
	Cloud Services	✅	✅
Reference Apps	AI NVR	✅
	Generative AI—Zero-Shot Detection With Prompts	✅
	Multi-Camera Tracking		✅
	Occupancy Analytics		✅
	Few-Shot Product Recognition		✅

Transform Your Workflow

Detection, Tracking, and Embedding

Detect, track, and generate embedding for objects within each camera stream. The microservice can handle multiple camera streams concurrently.

Multi-Camera Tracking

Track movement of objects across multiple cameras.

Behavior Analytics

Determine behaviors and anomalies in movement of objects.

Behavior Learning

Reveal and predict behavior patterns using clustering and deep learning. Learning is continuous.

Video Management and Storage

Discover, connect, and manage cameras. Also, manage video streaming and storage.