YUAN Powers Next-Generation Smart City NVR with NVIDIA Jetson Thor
Integrating high-performance AI computing, multi-channel PoE, high-speed networking, and local storage to transform city video systems from passive monitoring into real-time perception, analysis, and event response.
From Video Surveillance to Real-Time Urban Intelligence
As cities accelerate smart infrastructure deployment, traditional recording systems are evolving into intelligent platforms capable of understanding events in real time. The YUAN EDG6N0-S T5X NVR is powered by NVIDIA Jetson Thor, delivering up to 2070 FP4 TFLOPS of AI performance and 128GB LPDDR5X memory with 273 GB/s bandwidth. By integrating AI inference, video processing, and storage at the edge, it reduces the need to send large volumes of raw video to the cloud while enabling faster event response.
From Object Detection to Event Understanding
Designed for roads, stations, public spaces, campuses, and government facilities, the edge AI platform combines powerful GPU computing with multimodal AI / VLM technologies to move beyond recording people, vehicles, and objects toward understanding real-world situations.
• Event & Safety Detection: Identify traffic accidents, restricted-area intrusion, unattended objects, abnormal behavior, fires, and other incidents.
• Traffic & Flow Analysis: Monitor vehicle flow, pedestrian movement, and road congestion in real time.
• Faster Response & Management: Provide timely alerts and situational information to support faster decision-making and incident handling.
8-Port PoE for Simplified Multi-Camera Deployment
Eight built-in PoE ports transmit video and provide camera power over a single Ethernet cable, simplifying multi-camera deployment at intersections, stations, campuses, and other distributed smart city sites.
By integrating camera connectivity and AI computing in one system, the platform streamlines the path from video capture and AI analysis to event alerts, turning live video into actionable information at the edge.
100/75GbE High-Speed Networking from Edge to Control Center
As camera counts and video resolutions increase, network bandwidth becomes critical. The system provides QSFP28 connectivity at up to 100Gbps, along with a 5Gbps RJ45 port, for connection to central control centers and management platforms.
In a distributed architecture, edge nodes can perform AI analysis locally and transmit key alerts, metadata, and event video to the control center through high-speed networking: Camera → Edge AI → High-Speed Network → Control Center.
High-Density Video Encoding and Decoding
The 128GB configuration supports up to 10 channels of 4K60 or 82 channels of 1080p30 decoding, and up to 6 channels of 4K60 or 54 channels of 1080p30 encoding, with support for H.26X, AV1, and VP9.
A single platform can handle multi-channel video streaming, recording, decoding, and AI inference, reducing the complexity of deploying and maintaining multiple systems.
Local Storage for Critical Event Footage
The system provides one M.2 2280 M-Key NVMe slot for OS storage and two 2.5-inch SATA HDD bays for data storage. When AI detects an abnormal event, relevant footage can be stored locally while alerts, metadata, or selected clips are sent to the central platform.
This edge-first architecture reduces unnecessary upstream bandwidth while preserving critical footage before and after an event for review and investigation.
Flexible I/O for Scalable Smart City Edge AI
EDG6N0-S T5X NVR provides M.2 M/E/B Key, USB 3.2 Type-A / Type-C, HDMI 2.0, and DisplayPort 1.4 interfaces, enabling flexible expansion of communications, storage, and peripheral devices based on project requirements.
Combined with NVIDIA Jetson Thor AI computing, 8-port PoE, high-density video processing, high-speed networking, and local storage, the platform advances traditional passive recording into a scalable Edge AI system with real-time perception, analysis, and event response for transportation, public safety, and urban infrastructure.