Let’s Build Smarter, Safer

Spaces with AI
Powered by NVIDIA Metropolis,
NVIDIA Jetson, and NVIDIA EGX Platform
The need to automate and improve operational NVIDIA Metropolis: Powering the World’s
efficiency and safety in our physical spaces has never Smartest Spaces
been greater. Approximately 1 billion video cameras—
In the pages that follow, we’ll explore real-world
the ultimate Internet of Things (IoT) sensors—have
examples of how AI enabled video analytics based
been deployed throughout the world’s cities and
on the NVIDIA Metropolis application framework
spaces to help us live better and safer. Optimizing
running on the NVIDIA EGX™ platform solve complex
AI-enabled video analytics is critical for frictionless
operational and safety problems across a broad range
retail, streamlined inventory management, traffic
of industries. NVIDIA Metropolis and its extensive
engineering in smart cities, optical inspection on
partner ecosystem help to bring powerful AI-based
factory floors, patient care in healthcare facilities,
video analytics applications to market faster.
and more.
Advantech Industrial AI Solutions
NVIDIA Jetson
AI Camera

at the Edge
NVIDIA Jetson
Advantech Edge AI Box

Advantech Industrial AI Solutions Advantech platforms are now validated with NVIDIA
at the Edge technologies that include NVIDIA Jetson and GPUs,
and are ready for NVIDIA Metropolis, NVIDIA- NVIDIA RTX
Advantech provides comprehensive edge AI
Certified System, and NVIDIA Clara. By collaborating GPU IPC
platforms embedded with NVIDIA technologies—
with NVIDIA computing technologies and ecosystem
including AI camera, edge AI computing platforms
partners, Advantech enables industries to bring
(MIC-AI series), AI video systems (MIC-IVAs), GPU
AI to the edge.
IPC, and GPU servers. These AI platforms are widely
used in automation manufacturing, Smart Cities
Industries
applications, smart agriculture and medical imaging.
> Industrial automation
With 30 years of expertise in industrial PC,
Advantech provides industrial design solutions > Transportation NVIDIA-Certified
that are capable of enduring harsh environments GPU Server
> Safe and security
and feature flexible I/O, expansion, and interfaces.
These interfaces let customers adapt to different > Medical imaging and telemedicine
application requirements, and enable the use of More product information
> Warehouse and logistics
wireless communication modules (4G LTE, 5G, or RS- http://bit.ly/AILandingPage
485) and industrial interfaces. > Agriculture
Contact us:
> Retail and infotainment http://bit.ly/Advantech_Contact
Traffic Driving Smoother, More Sustainable
Traffic for 4.5 Million Drivers
detect and classify road users, including cars, buses,
trucks, bicycles, pedestrians, and even emergency

Management
vehicles in any lighting or weather conditions. The
Arizona’s Maricopa County Department of
processed data is then used to optimize traffic-light
Transportation (MCDOT) had a serious congestion
management with edge-based computing, saving
issue—managing the nation’s 14th most traffic-heavy
bandwidth and lowering latency. City engineers can
NoTraffic region. NoTraffic, an NVIDIA Metropolis partner, gave
them the answers they needed with an AI-based
view the data using custom dashboards for real-
time data analysis that enables capabilities such as
traffic management platform that optimizes flow
collision prediction and prioritizing certain road users.
based on real-time utilization and demand.
MCDOT realized significant efficiency improvements,
AI sensor units are installed at each intersection,
with more to come. An estimated 22,607 metric tons
providing road-user detection and classification
of greenhouse gases could be avoided annually—
using a fusion of machine vision and radar. The
like taking 4,915 vehicles off the roads. Over $50
sensors use the NVIDIA Jetson™ edge AI devices
million could be saved annually through costs
and NVIDIA Metropolis AI framework to process
prevented (vehicle wear and tear, fuel savings, and
video of the intersection traffic. NoTraffic’s units
maintenance). And an estimated 3,957 months of
vehicle delay time could be eliminated each year.
Roadside the future as the district’s needs continued to grow.

The DOT needed to find a video monitoring solution

Jetson models are validated to ensure compatibility
with Jetson and are ideal for video capture, recording,

Infrastructure
and streaming with multiple I/O support. With an
that didn’t fit into the traditional method of a loop-
NVIDIA Metropolis, MIC Jetson products also offer
detection underground system. An underground
fast AI deployment with AI function block.
system setup would have cost the district extensive

U.S. Department of construction time, along with increased expenditures

and maintenance efforts. Advantech instead
Featuring strict validation to ensure thermal,
mechanical, and electrical compatibility—plus industrial-
Transportation suggested an above-ground system that uses a grade anti-vibration, high temperature operation
video camera to process data and provide reliable, capabilities, and a modular, compact-sized design.
critical information to the district’s traffic controller, Advantech systems can run perfectly at the roadside.
Building a Custom, Future-Proof AI while avoiding unnecessary costs and maintenance.
Edge Computing Solution for Smart City The collected data from video cameras and sensors
Applications and Roadside Infrastructure would allow the district and traffic control centers
A U.S. department of transportation (DOT) needed to monitor and make proper decisions to help with
help from Advantech to create a powerful and robust traffic congestion, traffic flow, operational efficiency,
Artificial Intelligence (AI) Video Monitoring system etc. Additionally, an intelligent AI video system with
to help improve road infrastructure and traffic edge data analytics can help provide more predictive
management in its districts. The goal of the final travel time and help to effectively relieve traffic
system was to enable safer transportation on the intersection issues.
streets and help solve traffic congestion by tracking The Advantech MIC Jetson series, powered by the
objects and delivering accurate actuation and data NVIDIA® Jetson™ platform, offers all the performance
collection. Additional requirements included the of a GPU workstation in an embedded module. MIC
ability to support 5G and smart city applications in
AI Smart Factory In the past, manufacturing plant operators manually
recorded machine control panel data. This production
GoodLinker and Advantech designed the Screen Data
Extractor (SDE) to address data collection issues
yield and product quality data was transferred to a by integrating software and hardware. Advantech’s
computer where it became difficult to retrieve. The MIC-710AIX is industrial-grade edge AI computer
GoodLinker lack of internet connection capabilities, or limitations embedded with NVIDIA Jetson platform. MIC-710AIX
caused by independent/closed networks, necessitated leverages the GoodLinker OCR recognition, yielding
this stressful and inefficient process. Fortunately, an excellent SDE solution. This system exploits VGA/
“Screen Data Extractor (SDE)” Aimed at AI industry 4.0 renders this approach obsolete. HMI data output to ocular-recognize and record data
Smart Factory Transformation presented on screen. SDE is an external device; as
“Machine networking” is the first step to smart
such, data collection can be completed with minimal
factory transformation. Only by integrating machine
impact on the production plan and production line.
data can one proceed to the next step – data analysis.
This means locating problems within available data Advantech’s MIC-710AIX AI inference system
and making improvements. Relying on manual labor features a compact, all-in-one design that offers
requires the help of a network or system. Despite NVIDIA Jetson GPU computing support. It leverages
this, the solution used by most manufacturers is a modular interface design and offers diversified
either to update old machinery or purchase new image output methods to meet the requirements
models. Both solutions are potentially costly and diverse equipment. GoodLinker has used MIC-710AIX
impact production line efficiency. Installing sensors to implement SDE in factories; where it adapts to
is a relatively low-cost method when applicable. different output I/O without customization. The
First, certain expensive large-scale machines are not whole process is completed without compromising
suitable for self-installing sensors. Second, sensors on-site images.
can only retrieve physical data related to operating
Download Full Case: https://bit.ly/Adv_NV_SDE
rate, number of operations, and vibration. Sensor
limitations hinder the collection of non-physical data.
AI Automation Manufacturing
Smasoft

Deploying AI Inspection for Heat Expansion Slot Compact Fanless System MIC-770 30 seconds to conduct three types of inspections
Sink Factory with NVIDIA RTX and features a built-in NVIDIA® on each heat sink. Today, only one inspector is
Jetson AGX Xavier™ MIC-730AI AI Inference System. required to conduct final re-inspection and it only
In the past, most factory components were visually
takes 4 seconds to complete the task. Therefore, the
inspected by employees. Although prior education The heat sink factory originally had three major
introduction of the solution not only reduces labor
training was provided for inspectors, inspection inspection stations, in which multiple inspectors
costs but also improves inspection efficiency.
based on individual perception made it difficult to conducted size, thickness and appearance
integrate inspection standards. Moreover, visual inspections. Smasoft’s current automatic software
fatigue, weariness and staff turnover result in high development platform integrates modularized
missed inspection rates for production lines. visual inspection, motion control, I/O control, AI
defect inspection tools, AI training and other
After the defective products are released into the
functions, creating a heat sink inspection machine
market, the complaint rate has skyrocketed.
that can conduct three types of inspections while
Smasoft deliver a solution which combines rule- continuously training AI deep learning models to
based machine vision and AI visual identification is further lower the missed inspection rate and enhance
not only the perfect answer to a factory’s problems inspection quality.
but also contributes to the introduction of automatic
The client factory expected the new system to have
inspection systems, that help manufacturing
90% accuracy rate, but the Smasoft and Advantech’s
industry shift from human visual inspection to AI
solution achieved an accuracy rate of up to 97%,
inspection. Smasoft‘s solution play an intricate role in
far exceeding the client’s requirements. In the past,
optimizing the AI system of the heat sink inspection
there were ten inspectors to a production line. It took
machine. Smasoft adopts the Advantech i-Modules
Surgical Intelligence Platform
Theator

Medical-Grade Computing Facilitates effectively, theator needed a medical-grade Edge AI

Healthcare AI computer with substantial video-processing power
and the ability to capture video of up to 4K quality
With the current growth in population and life
and execute AI algorithms.
expectancy worldwide, the cost and workloads of
health professionals is expected to increase year on To satisfy their requirements, Advantech provided its
year. However, the emergence of artificial intelligence USM-500 medical-grade Edge AI computer. USM-500
(AI) facilitates innovations that can transform features an innovative design with multiple expansion
healthcare and overcome the challenges currently options for easy configuration without corresponding
faced by healthcare practitioners. drivers, components, or heat dissipation solutions.
For this project, USM-500 was equipped with two
A surgical platform developer, theator, is aiming to
video capture cards that support both 4K and full HD
integrate AI into everyday surgical practice. Together,
resolution and an NVIDIA RTX-4000 graphics card.
a team of leading surgeons, computer vision
These cards support the capturing of large video
experts, clinical specialists, and software developers
files and the execution of AI algorithms and analysis
have developed a surgical intelligence platform.
in real time. The collaboration between theator and
Powered by computer vision, this platform combines
Advantech is expected to take AI-based computing
sophisticated AI technologies that aggregate
applications to a new level by harnessing the power
data from thousands of surgical videos to provide
of surgical data to enhance and personalize
surgeons with a clearer perspective and more precise
patient care.
insights. To operate this surgical intelligence platform

Image credit: Computational Pathology Group

of Radboud University Medical Centre
Transit Hub Management
Assaia

Improving Airline Efficiency for 46.5 to implement Assaia’s Turnaround Control, installing
Million Passengers—Powered by AI 28 cameras at 14 stands. Video streams from the
cameras are analyzed on-premise using NVIDIA®
When a commercial flight lands and taxis to the gate,
Jetson AGX Xavier™ modules, giving the airport full
the choreographed dance of cleaning, restocking,
visibility of all turnaround activities in real time. The
and servicing a plane begins, known as the
system uses AI algorithms to turn the video streams
‘turnaround’. Accomplishing this faster means less
into structured data, including timestamps for
wasted time with an aircraft sitting on the tarmac,
turnaround events and safety-related data. Airport
less waiting around for customers, and more profits
staff and vendors can then use this data to increase
for the airline, London Gatwick Airport (LGW), the
on-time-performance, enhance safety, and
busiest single-runway airport in the world, tapped
decrease costs.
into the power of AI to make every second count.
LGW can now get more passengers onto more flights
Assaia is a Zurich-based startup that uses image
each day with lower operational costs, an accelerated
recognition algorithms to process video and offer
move towards sustainability, and a safer environment
predictive analysis to better manage turnaround
for airport workers and airline crews and passengers.
operations. LGW was the first airport in the world
Parking Operations and even navigate drivers to empty parking spots.
Camera streams are analyzed in real-time by a server
powered by the NVIDIA EGX platform and NVIDIA T4
GPUs using an algorithm trained to detect whether
DataFromSky individual parking spaces are free or occupied.
A browser-based interface also makes system
administration easy, and mobile apps connect drivers
Optimizing the Daily Commute Through with payment, navigation services, and notifications
More Intelligent Parking when their allotted parking time is about to expire.
For many commuters, the daily journey involves The system can even interface with local law
driving to the train station, parking, then catching enforcement agencies to share license plate numbers
the train. Sounds easy enough, but a lack of parking and other information should a crime occur in the lot.
spots where they’re most needed often leads to
Køge Nord Station can now offer commuters the
frustration and wasted time, and can even cause
ideal parking solution. Each camera monitors up
missed trains. The new Køge Nord Station, near the
to 400 parking spots, converting video streams
Danish capital of Copenhagen, took on this challenge
into valuable insights in real-time using the NVIDIA
using the power of AI.
DeepStream SDK. The system delivers new data and
An innovative new solution from NVIDIA Metropolis capabilities while being much cheaper and easier
partner DataFromSky uses cameras to monitor to implement than in-ground sensors. AI helps the
parking lots for occupancy, enable mobile payments, station run more smoothly and save money while
making every commute a little less painful.
Hospital Operations
Artisight

Creating a Safer Environment—and installed additional thermal cameras linked to its

Bottom Line—for Today’s Hospitals network at 104 entrances across the health system
network. using AI to detect roughly 50 cases of fever
Keeping today’s hospitals running smoothly and
among the 20,000 people coming through the doors
safely is more challenging than ever before.
each day. Compared to conventional temperature
NVIDIA Metropolis partner, Artisight, uses AI to probes, the camera system has shortened waiting
improve productivity in the operating room and help lines and reduced the staff needed at each door
hospitals improve their efficiency. Their applications by 75%.
generate HIPAA-compliant data from video cameras
Today, they’ve since deployed across 37 hospitals
and wireless devices installed strategically around
and ambulatory surgery centers, using computer
the facility, providing insights on workflows and
vision and conversational AI to improve operating
patient locations to optimize how care is provided.
rooms efficiency, increase patients’ satisfaction, and
Artisight had a network of 2000 cameras and strengthen the hospitals’ bottom line. As a result,
other sensors deployed across Northwestern’s ten nurses are going into rooms less—thereby reducing
hospitals running on the NVIDIA EGX platform. Then overall exposure — and consumption of protective
the pandemic hit, and the hospital asked for help gear is down. It’s all about safety and efficiency—
screening people coming in for COVID-19. Artisight powered by AI.
Workplace Safety
Helin

Making Dangerous Job Sites Safer—Even and alerts them before an accident occurs. It also Jetson edge AI platform powers the solution, which
at Sea lets clients examine how they work and determine offers unbeatable performance and energy efficiency
if portions of jobs can be done with tools and in a tiny form factor.
Oil rigs are some of the most dangerous job sites
machinery, instead of additional human workers.
in the world. But Dutch safety company Helin Data Today, the model pinpoints people’s locations within
That’s an opportunity to be more efficient and safer,
is making offshore rigs safer—and more efficient— 18 inches and sends an alert that can run a complete
given one less person potentially standing in an
with AI. sequence in under 250 milliseconds if someone is
unsafe position.
in an unsafe area. Real-time analytics at the edge
Helin’s application uses a network of cameras and
Oil rigs don’t have internet access, so Helin had to is critical to make this happen. Every real alarm
sensors feeding into AI-powered video analytics
design its system to handle all of its data capture can save a life and every false alarm serves as an
algorithms based on the NVIDIA Metropolis stack.
and processing using on-site hardware. The NVIDIA opportunity to refine the model.
The system locates rig workers in unsafe spots
Campus
Management
Icetana
Discovering New Insights to Keep on the carpet. When a guard went to investigate, he
Campuses Safe During the Covid-19 found a major water leak. Without icetana’s AI video
Shutdown analytics, the leak would have gone on undetected,
Mount Royal University in Calgary, Alberta, Canada as that part of campus had become vacant during
had been using icetana’s public safety system for the shutdown. Instead, campus staff were able to
about a year when the COVID-19 pandemic took contain the damage and cost of the leak almost
hold. The system applies real-time GPU-accelerated instantaneously.
AI analysis to more than 250 campus cameras, Icetana’s application uses the power of the NVIDIA
differentiates routine motion from unusual EGX platform to analyze real-time video footage
activity, and sends real-time notifications to and create immediate alerts for any abnormal
operations teams. activity, alleviating the need to have staff constantly
When campus access was restricted for COVID-19 monitoring every camera.
safety reasons, the AI system adapted to the reduced Since installing the system, Mount Royal operations
movement of students and staff. As it turned have reported an average of four actionable
out, not all unusual activity on campus involved events each week that would have gone unnoticed
humans. icetana’s system reported a mysterious and unreported. This is a great example of AI
anomaly along a hallway inside one of Mount Royal’s solving operations and safety problems in a
residential buildings. Control room staff were able to single application.
review the video feed and spot a dark patch forming
Airport Safety Managing the Growing Pains of
Indonesia’s Largest Airport
monitoring, passenger counting, baggage handling
violations, perimeter protection, abandoned baggage,
and loitering. It also lets operators track abandoned
Jakarta’s Soekarno-hatta International Airport (CGK)
vehicles—one of the highest threats at any airport.
Ipsotek was facing a growing challenge. With three terminals
serving more than 66 million passengers a year, and
The AI-enabled application can detect when a driver
abandons a vehicle, triggering a real-time alarm and
another 44 million anticipated by 2024, it was crucial
auto-tracking the driver using pan-tilt-zoom (PTZ)
for the airport to accurately identify and manage
cameras.
incidents as quickly and efficiently as possible to
minimize disruptions to passengers and airside Ipsotek’s application generates real-time and
operations. archived data for valuable business and health and
safety insights. This includes people-counting over
Beginning in 2015, Ipsotek set to work deploying an
time, crowd density with heatmaps, occupancy
AI-powered solution to analyze video camera streams
levels and distribution, social distancing, and face-
for a number of operational use cases throughout
mask detection. It all makes for a smoother, safer
the terminals and is powered by the NVIDIA EGX
environment for both workers and travelers.
platform. The capabilities deployed included exit lane
Crowd Safety
Awiros

Keeping a Big Crowd Safe—10 Million The application was deployed on cameras positioned
People Big at several key entry-exit points of the festival site, as
well as on a drone camera that provided an aerial view
Medaram Jathara is the biggest tribal festival in Asia,
of the venue. Video feeds were analyzed in real time
drawing nearly 10 million devotees to Telangana, a
using the NVIDIA DeepStream SDK and powered by
state in southern India, every year. Managing a crowd
the NVIDIA EGX platform. Each GPU-enabled system
this size takes more than extraordinary planning and
analyzed video streams from several live cameras
coordination. It takes powerful technology.
simultaneously, keeping the system footprint to a
The state authority responsible for event safety minimum and greatly reducing cost.
used the Awiros Crowd Estimation application to
Officials were alerted when the crowd density in a
measure crowd density at several locations during
region crossed a specific safety threshold or when
the 2020 festival. This first-of-its-kind tool uses AI
the rate of change in crowd density changed rapidly
to estimate crowd densities in real time using video
and could act quickly—keeping attendees safe and
cameras, letting authorities detect crowd size and
the festival running smoothly.
track variations in area-specific density over time.
This gives them a head-start on taking preventive
measures and avoiding dangerous situations such
as stampedes.
Retail Operations
AiFi

Fast, Easy, Stress-Free Shopping—No The “Nanostores” run on the O.A.S.I.S. Autonomous
Cashier Required Store Platform, which includes seamless multi-
camera, multi-person tracking, real-time product
Great news for anyone who loves shopping but hates
recognition, and live inventory tracking to deliver an
lines. The convenience and speed of checkout-free
ultra-fast, checkout-free shopping experience. The
shopping is becoming increasingly accessible in more
platform also provides shopper-behavior analytics
places—from airports, to gas stations, and even to
and sales and brands analytics, and supports flexible
the Indy 500 auto race track—thanks to AiFi’s AI-
payment options.
powered automated retail solutions. And there are
many more sites on the way. Full autonomy allows Nanostores to operate around
the clock, 365 days a year, with no additional
AiFi uses a combination of AI, edge computing, and
overhead. They can also be deployed, moved, and
IoT fusion technology to create an autonomous retail
redeployed in a new location with minimal downtime
environment. Shoppers tap a (contactless) debit card
or cost, making them perfect for pop-up stores,
or compatible wearable device at the door to gain
festivals, sporting events, and college campuses.
entry, get the products they want, and simply leave.
In addition to Nanostore format, AiFi’s cashierless
The items are automatically registered—using AI-
checkout technology has also been deployed in a
enabled video analytics powered by the NVIDIA EGX
larger convenience store format across the globe.
platform—and paid for, with receipts transmitted
wirelessly to shoppers.
Quality Control
Data Monsters

Reducing Packaging Defects at the variability, condensation, vibration, can placements,

World’s Largest Brewer and frequently changing package design, not to
mention Internet bandwidth. But it’s a challenge well-
One of the world’s largest brewers is tackling the
suited for AI.
most challenging supply chain inspection problems–
detecting defects on beverages packaged in Data Monsters, an NVIDIA Metropolis partner,
aluminum cans being processed on high-velocity developed an AI application for the brewing
conveyor belts. There are about 50B aluminum cans company that used GPU-accelerated, real-time
processed each year in the US on high-speed belts AI vision inference and edge re-training capability
such as the ones this brewing company runs. It’s with specifically designed self-supervised learning
a challenging problem given the line speed, defect algorithms in-the-loop.
Package Sorting Using AI to Improve Delivery and
Processing on Billions of Packages
Complex feature data is extracted on approximately
20 million packages per day to enable fast package
search, along with maintaining package safety.
The USPS operates the world’s highest volume
Accelerated edge servers powered by NVIDIA EGX
United States Postal logistics operation, processing and delivering 146
billion pieces of mail each year, including more than
are deployed at all 192 USPS package sorting sites

Service (USPS) 6 billion packages. USPS is using AI and NVIDIA

running real-time video analytics with NVIDIA Triton™
Inference Server and NGC containers.
Metropolis to process package data 10X faster and
with higher accuracy—greatly improving package-
sorting site productivity.
Waste Material Reducing Waste Going into Landfills by
Enhancing Recycling Capabilities
analytics. These - deliver high throughput and
accuracy, as well as the ability to fit the compute-

Processing
intensive algorithms into space and power-
Global consumption is accelerating, with $2 trillion
constrained industrial environments using NVIDIA
worth of materials thrown away each year globally,
Jetson Xavier AGX. Recycleye’s models can run at
of which only 8% is recycled. Finding new ways to
60FPS, which means higher-capacity sorting thanks
Recycleye increase worldwide recycling levels is necessary to
speed the transition towards a circular economy.
to faster conveyor belt speeds.

Recycleye’s technologies exceed human performance

Recycleye deploys AI-powered automated waste
in identifying, classifying and sorting individual items
analysis and sorting systems to improve material
on waste streams, with over 95% accuracy 24/7. At a
recycling facility efficiency, cost, and speed. Their
fraction of the cost of existing systems, Recycleye’s
computer vision system detects and classifies items
solutions have been installed across the UK and
in the waste sorting process by material and object.
France, enabling recycling facilities to tackle more of
Then, Recycleye uses the NVIDIA DeepStream SDK
the world’s waste.
and TAO Toolkit to deploy highly optimized video
Advantech Computing Platforms with
NVIDIA Solutions
MIC-730AI MIC-720AI MIC-710AILX/MIC-710AIL MIC-710AI/MIC-710AIX
AI Inference System on AI Inference System on AI Inference System on NVIDIA AI Inference System on
NVIDIA Jetson AGX Xavier™ NVIDIA Jetson Tegra® X2 Jetson Xavier™ NX/NVIDIA NVIDIA Jetson Nano™/
Jetson Nano™ NVIDIA Jetson Xavier™ NX
Edge AI Solutions
With NVIDIA Jetson™ MIC-715 MIC-733-AO MIC-737-AO
EPC-R7000
AI Inference System on NVIDIA Jetson™ TX2 Edge AI Inference System on AI Inference System on
NVIDIA Jetson Xavier™ NX AI Inference System NVIDIA Jetson AGX Orin™ NVIDIA Jetson AGX Orin™

MIC-770 V2 + MIC-75G20 ACP-2020G + AIMB-787 IPC-7130 + AIMB-787 MIC-770 + MIC-75G30 ACP-4020 + AIMB-787
Industrial Computers NVIDIA PCIe GPU up to NVIDIA PCIe GPU up to NVIDIA PCIe GPU up to (2x) NVIDIA PCIe GPU NVIDIA PCIe GPU
With NVIDIA GPUs 350W TDP 350W TDP 250W TDP up to 250W TDP up to 250W TD

MIC-730IVA MIC-710IVX
AI Video Analytics 8ch AI Video System on 8ch AI Video System on
With NVIDIA Jetson™ NVIDIA Jetson AGX Xavier™ NVIDIA Jetson Xavier™ NX

SKY-6100 SKY-6200 SKY-6400 SKY-6420

Rackmount Servers 1U 5 NVIDIA GPU Cards 2U 4 NVIDIA GPU Cards 4U 4 NVIDIA GPU Cards 4U 10 NVIDIA GPU Cards
With NVIDIA GPUs AI Inference Server AI Hybrid Server AI Training Server High-Density AI
Training Server
Work from Anywhere with NVIDIA
Advantech has developed a series of AI hardware solutions using NVIDIA GPUs to transform our spaces.
Advantech’s partnership with NVIDIA is taking huge steps forward in making AI a reality for a wide range of
industries such as healthcare, manufacturing, transportation, telecom, and more.

Ready to get started?

Learn more about NVIDIA Metropolis, NVIDIA Jetson, and
NVIDIA EGX platform.
Contact us: [email protected] | 886 2 2792-7818 ext.1782
© 2022 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, Jetson AGX Xavier, NVIDIA EGX, NVIDIA Jetson,
and NVIDIA Triton are trademarks and/ or registered trademarks of NVIDIA Corporation in the U.S. and other countries.
All other trademarks and copyrights are the property of their respective owners. 2469150. Sep22