Overview
Metro AI Suite is an application framework with libraries, tools, and reference implementations, to enable partners to create Vision AI and Gen AI solutions in the Safety and Security, Transportation, and Critical Infrastructure markets.
What's New
Announcing Metro AI Suite 2026.0
Metro AI Suite expands agentic and physical AI capabilities by strengthening scalable video analytics, system evaluation, and real‑world edge AI deployment on Intel® platforms. This release enhances Deep Learning Streamer for pipeline optimization, improves Visual Pipeline and Platform Evaluation Tool usability for benchmarking, advances SceneScape™ with object tracking that scales to 1,000 concurrent objects, and includes new solutions for transportation and cities.
Learn MoreIntroducing Intel® Core™ Ultra Series 3 Processors
Intel® Core™ Ultra Series 3 processors bring deterministic, real-time performance to Metro and Smart City deployments — powering multi-camera traffic analytics, urban infrastructure monitoring, and intelligent edge gateways through heterogeneous compute.
Learn MoreAI Suite Benefits
Metro AI Suite provides a range of tools, components, and programs to help solution builders, from device manufacturers to software developers, through the end-to-end process of solution creation on Intel at the edge. This covers three stages:
Validate and Size AI Systems
Identify performance requirements and test hardware configurations on partner-validated AI systems based on x86 architecture
Demonstrate AI Acceleration and Real-Time Performance
Prove viability running real-world AI workloads to balance the right compute, power requirements, and form factors to suit diverse Metro applications.
Develop AI applications faster
Use reference implementations as blueprints for developing AI solutions optimized for Intel® hardware. Or migrate existing systems to Intel® hardware without extensive refactoring.
Technology & Performance
Explore the advanced Vision AI and Gen AI capabilities and video analytics pipelines that enable the Metro AI Suite to deliver proven, repeatable performance on Intel processors at the edge.
Vision AI and Gen AI Capabilities
Discover the underlying technologies that enable Metro AI Suite’s smart city, traffic, and public safety use cases, and benchmarking capabilities to help validate performance.
Vision-Language Models (VLM)
Metro AI Suite integrates VLM capabilities to understand, process, and respond to video-, LiDAR-, and image-based data inputs, enabling complex reasoning with multimodal fusion in the Smart NVR and Video Search and Summarization reference applications.
Physical AI
Enable Physical AI with Scenescape by fusing multi-sensor inputs into a live 4D digital twin that models objects, movement, and environmental context over time supporting situational awareness, forecasting, and intelligent action in real-world environments.
AI Model Development
Create, train, and fine tune computer Vision AI models easily and quickly without technical expertise. Upload curated images and start annotating in a graphical user interface to teach the model which objects you want to recognize and Geti trains and optimizes models for precision and accuracy on Intel silicon.
Media Analytics
Deep Learning Streamer (DL Streamer) is a GStreamer-based framework for constructing real-time video analytics pipelines with AI inference on Intel® platforms. Together with the DL Streamer Pipeline Server, it enables scalable deployment and management of performance-optimized media processing pipelines at the edge.
Benchmarking
Use Visual Pipeline and Platform Evaluation Tool (ViPPET) to right-size hardware selection and visualize performance across Metro AI workloads to analyze throughput, CPU and GPU usage, and memory utilization to tune the right balance of cost, performance, and efficiency.
Open Edge Ecosystem
Intel’s empowers partners to seamlessly integrate AI solutions for smart cities, transportation, and security to extend x86 architectures for fast decision making, improved efficiency, and reduced latency through an open, collaborative approach ensure performance, scalability, and lower overall cost.
Verified AI Edge Systems
Intel partners with OEMs and ODMs to benchmark and verify hardware solutions optimized to meet heterogeneous computing requirements for IoT and AI-driven metro and public safety deployments to ensure optimized hardware and software integration across a range of power levels, sizes and performance options.
Metro SDK Manager
Metro SDK Manager eliminates the complexity of manually configuring SDK combinations by providing automated dependency resolution, version compatibility checks, and integrated toolchain setup for the Metro Vision AI SDK, Metro Gen AI SDK, and Visual AI Demo Kit.
Evaluation, Qualification, and Benchmarking
Build and deploy edge AI applications faster with tested, verified, qualified AI Systems, running real-world edge AI workloads, to ensure predictable performance, long-term reliability, and streamlined, scalable deployment.
Intel® ESQ for Intel® AI Edge Systems
This program enables solution builders to test and qualify AI systems running industry AI workloads to prove industrial reliability and viability for powering edge AI innovation.
Metro AI Suite Performance Blueprints
Scaling Vision AI and Gen AI across distributed edge deployments and cloud are a constant tradeoff of efficiency, precision, performance, and cost. Adapting the Metro AI SDK and Suites and Intel systems to industry workloads provides a happy path to selecting a system optimized for the right AI workloads, deployment scenarios, and cost-of-ownership goals.
Visual Pipeline and Platform Evaluation Tool
Analyze throughput, CPU usage, and GPU usage with an open-source tool that enables custom performance benchmarking and KPIs to validate AI inference pipelines to measure latency and resource utilization.
AI Inference and Media Pipeline Benchmarks
Visualize what’s possible when comparing generation-to-generation of Intel processors or Intel vs competitors’ processors to understand the differences.
Intel® Core™ Ultra Processor Series 3 End-to-end AI and Video Pipeline Benchmarks
Measure performance gains of Intel® Core™ Ultra Processor Series 3 over previous generations. (CNDA required)
Competitive End-to-end AI and Video Analytics Pipeline Benchmarks
Compare performance differences between Intel® Core™ Processors and competitors’ solutions. (CNDA required)
Sample Applications
The Metro AI Suite features a collection of open-source sample applications and tutorials that demonstrate AI use cases for smart cities, traffic, infrastructure, and public safety.
Loitering Detection
Monitor and identify suspicious behavior in real time and proactively address potential security threats with fine-grained detection and streaming video analytics.
Live Video Alert Agent
Deploy AI-powered video alerting using OpenVINO™ VLMs to process real-time streaming protocol (RTSP) streams to generate real-time alerts from natural language prompts and monitor them on a unified dashboard for real time video analytics and safety monitoring. Part of the Live Video Analysis framework.
Smart NVR
Use GenAI-powered vision analytics to transform a traditional NVR into a Smart NVR, converting conventional network video recorders into intelligent, context-aware systems, unlocking advanced insights and automation at the edge.
Video Processing for NVR
Process video analytics and transcoding, configure concurrent video processing, including video decode, post-processing, and concurrent display, utilizing the integrated GPU on Intel® Core™ Ultra Processors.
Smart Route Planning Agent
Finds optimum routes in real-time with AI Agent that fetches and analyzes live feeds from traffic intersections.
Smart Intersection Agent
Monitor intersections in real-time with directional traffic density analysis and Vision Language Model (VLM) powered traffic insights. Manage MQTT traffic data and camera images for intelligent traffic analysis.
Smart Intersection
Incorporate real-time, multi-camera scene analysis, and spatial intelligence into traffic management scenarios to understand intersection dynamics and enable pedestrian safety.
Smart Parking
Detect empty parking spaces in real-time with AI-driven video analytics using pre-trained deep learning models and video streams.
Sensor Fusion for Traffic Management
Integrating and reconciling data from multimodal sensor inputs, the Metro AI Suite uses frameworks and microservices to process raw sensor data into a unified model for actionable insights. Scenescape combines sensor data into dynamic virtual renderings of buildings, streets, and city grids to enhance monitoring with Digital Twins.
Deterministic Threat Detection
Set up and run this Time-Sensitive Networking (TSN) sample application for deterministic, low-latency delivery of AI-processed video and sensor data in a shared network, to protect critical workloads from critical network congestion.
Live Video Search
Adapt the Video Search and Summarization pipeline for semantic search on live camera streams, with video indexing with embeddings and timestamped camera metadata, to enable precise selection of camera, time range and free-text queries to retrieve, rank, and play clips with confidence scores. Part of the Live Video Analysis framework.
Live Video Captioning
Live video Captioning deploys AI-powered captioning for live video streams with Deep Learning Streamer (DL Streamer) and OpenVINO™ VLMs for real-time video analytics, accessibility enhancement, performance benchmarking, and intelligent surveillance. Part of the Live Video Analysis framework.
Visual Search Q&A
Inter a multi-modal search engine with a multi-modal visual Q&A chatbot that answers questions based on search results, and a graphical user interface to ask questions and scan results.
Image Based Video Search
Search live or recorded camera feeds in near real time to detect and retrieve objects of interest from large video datasets, matching objects with location, timestamp, and confidence score details using vector-based search.
Interactive Digital Avatar
Improve user engagement and enable better support with 2D and 3D digital avatars backed by an LLM server to provide real-time intelligent responses to user queries with conversational capabilities.
Insights and Resources
Learn more about how the Metro AI Suite can deliver optimal AI acceleration and media pipelines and real-time performance at the edge to enable smart AI for traffic intersections, public safety, and infrastructure related use cases.
Use these resources to configure Intel hardware and deploy the sample applications, AI libraries, and frameworks in the Metro AI Suite.
Metro AI Suite Developer Guide
Use this guide to get familiar with the Metro SDK, sample applications, blueprints, tools and programs to evaluate Metro AI workloads.
Intel Hardware Partner Guide
Remove the guesswork from selecting and scaling edge AI hardware systems with a guide to get started with sizing guidelines, benchmarks, best-known configurations, and materials to help device manufacturers connect with and engage the AI development community.
Investigate the many facets of the Metro AI Suite’s Vision AI sample applications and tutorials, as well as the Metro SDK, to get up and running quickly toward securely deploying and scaling AI at the edge.
Smart Intersection (Metro AI Suite | Open Edge Platform)
The new Smart Intersection sample application that analyzes streaming media analytics across multiple intersections in real time to monitor traffic flow and pedestrian safety.
Smart Parking (Metro AI Suite | Open Edge Platform)
An introduction to the Metro AI Smart Parking sample application that uses Vision AI and streaming pipelines to monitor parking space availability in real-time.
Build Vision AI Without Being an Expert
Geti lets you build vision AI without being an expert. Understand the parameters of building customized AI applications.
Model Selection Made Simple
Build real-time video analytics pipelines with DL Streamer using a GStreamer-based framework that enables efficient video processing and AI inference optimized for Intel® edge platforms.
Build a video analytics pipeline using DL Streamer
Build realtime video analytics pipelines with DL Streamer using a GStreamerbased framework that enables efficient video processing and AI inference optimized for Intel® edge platforms.
Detect faces and features with streaming media pipelines
Run concurrent Vision AI model pipelines to encode and process video as well as detect faces, understand Age, emotion, landmarks and optimize for fast streaming.
How AI Training Works – Deep Learning and Fine Tuning Explained
Learn how deep learning pipelines work for fine tuning AI models across many layers to develop neural networks and inferencing multimodal media.
From Sensor Data to Spatial Intelligence Demo
Discover how sensor data gathered from multiple inputs gets transformed into a 4D digital twin to provide spatial awareness for Metro AI applications.
Smart Intersection Solution Demo
Get under the hood of how the smart intersection sample application ingests and transforms smart camera and IP video into multimodal analytics pipelines to combine a multi-scene view of intersections in real time.
Learn about the latest Metro AI trends across security, smart cities, and infrastructure.
Integrated AI, Unlocked Potential
Read about how recent evolutions in video analytics and visual AI are enhancing next-generation security use cases.
Learn from Intel experts through tutorials that share practical guidance for working with solutions in real‑world scenarios.
Performance Optimization
Learn how to optimize AI system performance by using hardware‑aware software optimization techniques.
Migrate from Nvidia
Migrate existing AI workloads to Intel® platforms using practical guidance, compatibility considerations, and optimized software paths.
Application Security Enablement
Enable built-in security features to ensure data privacy with strong encryption protocols and blockchain. Learn how to safeguard infrastructure with advanced threat detection to mitigate vulnerabilities across distributed edge devices.
Developer Tools
Try out the Metro AI Suite with development kits and tools to help build, evaluate, and deploy Vision AI and Gen AI use cases with accelerated media processing and AI inference optimized on Intel silicon.
Metro AI Suite makes it easier to download and install core functionality with microservices, tools, and frameworks to get up and running.
Metro SDK Manager
Metro SDK Manager eliminates the complexity of manually configuring SDK combinations by providing automated dependency resolution, version compatibility checks, and integrated toolchain setup for the Metro Vision AI SDK, Metro Gen AI SDK, and Visual AI Demo Kit.
Metro Gen AI SDK
Installs Open Edge Platform tools and microservices to adapt retrieval-augmented generation (RAG) capabilities. Features a practical question-and-answer reference implementation.
Visual AI Demo Kit
Installs tools and libraries for creating an end-to-end demo app for visual AI. Includes tutorials on using these components together to create sample applications for smart parking, smart intersection, and other Vision AI use cases.
Metro Vision AI SDK
Installs tools and libraries to develop video analytics applications optimized for efficient hardware performance. Includes tutorials and examples for using this to build an object detection pipeline with multi-stream video processing, multimodal AI pipelines, and model inference optimization.
Build, evaluate, and deploy Vision AI and Gen AI powered applications with tools for advanced AI inferencing, generating streaming media pipelines, fine-tuning Vision AI models, and benchmarking performance.
Geti™
Create, train, and fine tune computer Vision AI models easily and quickly without technical expertise. Upload curated images and start annotating in a graphical user interface to teach the model which objects you want to recognize and Geti trains and optimizes models for precision and accuracy on Intel silicon.
Scenescape™
Integrate data from cameras, LiDAR, and other inputs across multiple scenes to create 4D spatial intelligence with near real-time actionable data showing a digital twin of a scene mapping objects, attributes and the environment, including time-based sensor fusion to track past events, current activities and future predictions based on scene data.
Visual Pipeline and Platform Evaluation Tool (ViPPET)
Right size hardware selection and visualize performance across Metro AI workloads to analyze throughput, CPU and GPU usage, and memory utilization to tune the right balance of cost, performance, and efficiency.
OpenVINO™ Toolkit
Transform models from popular frameworks to run on Intel-based silicon and optimize compression to quantize models for edge deployments, for Vision AI and Generative AI use cases.
AI Libraries and Microservices
Leverage the power of the Open Edge Platform’s modular libraries and microservices to enable and optimize edge AI capabilities and streamline development time and effort to build and deploy advanced Vision AI and Generative AI applications.
Deep Learning Streamer™
Explore GStreamer‑based framework and pipeline server for real‑time media analytics with AI inference, featuring performance tuning, motion‑triggered inference, and DeepSORT‑based object tracking.
