Service Platform
Accelerate field deployment of the Edge AI Box with a
service platform that enables remote operation, management, and operation.
- HOME
- Service Platform
EDGEMATRIX Service Platform
Various EDGEMATRIX platform features free AI application developers from implementing various non-AI functions required for system integration for deployment, allowing them to focus on AI development. Stream + Device Agent” inside the AI Box and the management function on the Cloud side work together to provide platform services.(* EDGEMATRIX Service Platform subscription is required.)
Overall system configuration(* In the case of a single Edge AI Box)
Learn more about the technology ▼
Major platform features: Major Function
Project Management
It registers and manages a large number of users using the platform (including management of permissions for configuration, operation, viewing, etc.) and manages a large number of Edge AI Boxes and their connected cameras.
Camera Control
It is possible to remotely control ONVIF-enabled cameras, including frame rate and other stream settings and PTZ camera angle of view (pan, tilt, zoom) operations.
High-speed inference processing
The G-Streamer-based video pipeline architecture enables high-speed inference processing. For developers of AI applications, we also offer a Toolkit with reference models for developing real-time video edge AI applications.
Install / modify / update AI applications
AI applications pre-registered in the cloud-side repository can be installed on the AI Box, and applications can be modified and updated.
Action rule
Actions are executed by pre-setting alerts issued by AI analysis results, periodic transmission of inference results, and event transmission.
Map view
View the map of the Edge AI Box and camera locations and other registration information.
Edge view
It is possible to monitor the camera's live video (monitor view) and view and download event-recorded and constantly-recorded video files according to the recording settings.
Peer to peer (P2P) communication and NAT crossing function
All communications, such as sending analysis results, edge views, and remote control of the Edge AI Box, are done via P2P and support the Network Address Trandlation (NAT) traversal feature, which allows access to private networks.
Security function
Security measures include two-step authentication for user access, Edge AI Box authentication, encryption of storage and communication data, and neural model protection.
Life/death and anomaly monitoring
In 24×365 operation, etc., the Device Agent detects dead or abnormal conditions of the Edge AI Box and cameras, and notifies the user in the event of an abnormality.
Screenshot of the Management Console
Management Console (Device Console)Manual▶
Login screen (2-step verification)
Project Settings Screen
Camera setting screen
Map View Screen
PTZ camera adjustment screen
Available on the Service Platform
Edge AI Box
Advance MK2
Limited Stock
NX Outdoor
Limited Stock
NX Indoor
Limited Stock
STD Outdoor
STD Indoor MK2
Limited Stock
STD Indoor
Limited Stock
Light
EDGEMATRIX PaaS offerings
Deploy customer-developed applications on the EDGEMATRIX platform
EDGEMATRIX can solve these problems!
AI system installed nationwide
now how to operate it?
The platform remotely operates cameras and Edge AI Box! Centralized management of “Video + AI” at multiple locations without going to the site.
Can Edge AI be deployed
in any environment, outdoor or indoor?
The Edge AI Box series offers a lineup from low-cost to high-performance models. For outdoor use, we also offer dustproof, waterproof, and lightning-proof products that can withstand harsh environments.
Want to put PoC results
to practical use at low cost
AI apps developed through PoC are sold and distributed through the AI Application Library! Reduce the cost of commercialization and monetize in a short period of time.
Developers focus on AI model development
Deployment utilizes EDGEMATRIX PaaS
The EDGEMATRIX Platform as a Service (PaaS: Edge AI Box in conjunction with our cloud) allows customers to develop their own AI applications using the EDGEMATRIX PaaS and deploy them in the field using the EDGEMATRIX EDGEMATRIX PaaS enables customers to deploy their own AI applications to the field using the above functions of the EDGEMATRIX platform service. By using EDGEMATRIX PaaS for demonstration experiments, PoC, and practical application of AI, you can deploy AI applications efficiently and in a short period of time.
Contact us for more information about using EDGEMATRIX PaaS.EDGEMATRIX Stream
"EDGEMATRIX Stream" Service Platform on Edge AI Box
Used when providing services with the Ready-made AI System and Custom-made AI System (not included in the Edge AI Box stand-alone sales model).EDGEMATRIX Streamの特長
AIVA
AIVA (Artificial Intelligence Video Analytics) inference algorithm, which has a wide range of applications among AI, is executed by a uniform pipeline process (Primary AI ➡ Tracking ➡ Secondary AI ➡ Post Processing). ➡ Tracking ➡ Secondary AI ➡ Post Processing
increased portability of a wide variety of video AI applications, enabling efficient and fast processing of video
Deploy
EDGEMATRIX Stream and Device Agent work with EDGEMATRIX Cloud to provide various remote management functions
full support for post-PoC field deployment, a headache for developers
various management and operational functions are available from the device console in the cloud, including software updates, NAT crossing function, live monitoring function including remote PTZ, recording function, and action rule settings.
Security
Robust security features that are essential for real-world operations come as standard▪️Secure boot with unique key for each AI Box
▪️A strong login password with 252 bits of entropy
▪️Recording files and AI apps are encrypted with LUKS and stored on the secondary drive
▪️Even if the EDGEMATRIX service is disguised as a fake service, communication will not take place if it cannot be confirmed as genuine.
▪️The AI Box unique key and LUKS encryption password are not stored on the device, so even if the AI Box is stolen, the key is not stolen.
Overview of EDGEMATRIX Stream
EDGEMATRIX Stream is EMI's core streaming engine and provides the following features:
・NVIDIA DeepStream・Works with GStreamer plugin
・Integrated with EMI's edge AI services
as shown in the figure on the right, EDGEMATRIX Stream relies on NVIDIA's DeepStream for performance. Furthermore, EDGEMATRIX Stream provides both performance and efficiency for edge AI development.
Pipeline processing using core technologies such as NVIDIA DeepStream, GStreamer, and Web RTC enables high performance and functionality in AI processing using embedded GPU modules, including high-speed processing, multi-camera support, and remote control.
By focusing on AIVA, we have standardized the architecture as shown in the diagram below to increase the portability of AI applications.
Software configuration of
Edge AI Box using EDGEMATRIX Stream
The diagram on the right shows the overall software configuration of the Edge AI Box for EDGEMATRIX services such as the Ready-made AI System.
It mainly consists of NVIDIA JetPack, NVIDIA DeepStream, and EDGEMATRIX Stream.
*However, the software configuration is different for the Edge AI Box - Advance.
EDGEMATRIX Stream Toolkit
Provides a Developer Toolkit
Developers use the Edge AI Box toolkit box to develop/port AI applications. The Toolkit Box is a type of AI Box that is used to build your own apps using valuable pre-trained model binaries and comes with a GUI toolkit application with templates so you can start developing immediately. We also provide GUI toolkit software that allows developers to create AI applications using their own pre-trained model binaries.
