Applications

Edge analytics for city video.

City cameras and roadside sensors can generate terabytes of footage each hour. SmartNVR, SmartVMS, and SmartFlow process that data at the edge and organize events for city operations.

Aerial view of a multilane road interchange at night

275: TOPS of edge compute on VPP6N0-MB AGX
3: SmartApps in one ecosystem: SmartNVR, SmartVMS, SmartFlow
4K+: surveillance resolutions processed at the edge

The challenge

The data arrives faster than anyone can watch it.

Surveillance cameras and traffic sensors continuously generate large volumes of video and event data. A control-room team cannot watch every feed, and manual review often happens after the event.

Higher-resolution streams increase processing and network requirements. Running inference near the cameras reduces the need to send every frame to the cloud.

Surveillance control room with a wall of live camera feeds

The architecture

SmartNVR at the edge, SmartVMS in the cloud.

The SmartNVR VPP6N0 AGX processes camera data on site for traffic and public-safety operations. SmartVMS sends events and selected streams to centralized cloud dashboards with live monitoring, map integration, and multi-site management.

SmartFlow adds people-flow analytics. All three products use the NexVDO SDK to connect edge inference, anomaly detection, and system management.

Black compact edge-computing system on a desk

The watchpoints

Four uses for edge video analytics.

For pedestrian and vehicle flow, models can classify vehicle types and estimate movement for operator review. For public-safety workflows, the system can be configured to flag events such as unauthorized entry or an abandoned package; operators define the review and escalation process.

Traffic and disaster-prevention models can also flag possible violations, flooding, or fire. Detection performance depends on camera coverage, weather, lighting, training data, thresholds, and site validation. Public deployments also require privacy, retention, access, bias, and human-review controls.

Pedestrians crossing a city street viewed with motion blur

In the field

Wildlife detection and industrial CCTV deployments.

COMJO's AIJO Site Monitor runs on the YUAN Pandora platform with NVIDIA Jetson Orin NX Super Mode. The deployed system supports wildlife alerts, including bear detection, for Japanese government-agency projects. Swappable modules add helmet-compliance and meter-reading workflows.

Fogsphere uses YUAN's NVIDIA Jetson Thor T5000-based EDG6N0-S T5X for CCTV analysis and PPE-event detection in industrial and urban environments. Operators remain responsible for reviewing alerts and applying site safety procedures.