Edge-first · Events, not video

We turn existing cameras into a city-wide intelligence network.

NEXUS connects CCTV and ANPR feeds at the edge, fuses observations across cameras, and delivers vehicle journeys and traffic analytics from one graph.

Works with your stack

CCTV & ANPREdge AIPostGISMapLibreEvent streamOpen APIs

0

Video frames sent upstream

4

Signals in identity fusion

100+

Cameras per deployment target

1

Graph for journeys & traffic

Platform

Three layers. One intelligence graph.

NEXUS Edge

Detect, track, and OCR at the camera.

Every feed becomes an edge node. Video stays local. Upstream traffic is structured JSON events only: plate, appearance, timestamp, direction.

  • YOLO-family detection
  • ByteTrack / BoT-SORT
  • PaddleOCR · ONNX / TensorRT
Learn more

NEXUS Graph

Associate vehicles across the city.

Four-signal fusion: plate, appearance, time, topology. Outputs are scored and explainable: MATCH, PROBABLE, UNRESOLVED, CONFLICT.

  • Cross-camera Re-ID
  • Spatial-temporal graph
  • Never forced identity
Learn more

NEXUS Console

Operate journeys and city traffic.

Map-first operator console: live sightings, plate search with explain panels, corridor analytics, and human-gated alerts — wired to the NEXUS API.

  • MapLibre replay
  • Operator chat
  • Human-in-the-loop alerts
Enter war room

01 · Observe

Every camera becomes a sensor.

NEXUS sits on top of CCTV and ANPR you already deploy. Edge nodes run detection, tracking, and plate OCR locally, emitting structured events, never raw video streams.

See capabilities

Corridor live view

02 · Connect

One graph links the entire corridor.

Observations fuse across cameras using plate, appearance, time, and topology. The graph powers both vehicle identity and city-level traffic metrics from the same data.

See capabilities

Vehicle path · graph animation

Mission

The most important vehicle intelligence decisions happen at the edge and in the graph that connects them.

Enter war room

Sectors

One platform. Multiple missions.

Find any vehicle. Explain every match.

Search by plate or appearance, replay full journeys on a map, and review alerts with transparent scoring: plate, Re-ID, time, and corridor topology.

Includes

  • →Journey replay
  • →Explainable MATCH / PROBABLE
  • →Human ack before dispatch

Journey replay

Capabilities

Built for operators who need answers, not raw feeds.

Try the console demo

Cross-camera association

Plate, appearance, time, and topology, scored, never forced.

CameraSignalStatus
C03 · 09:41MH12AB1234MATCH
C07 · 09:46partial platePROBABLE
C12 · 09:53white sedanMATCH
C15 · 10:02MH12AB1234MATCH

Explainable operator search

Natural-language queries with journey replay and match rationale.

> show plate MH12 since 09:00
4 sightings · C03→C07→C12→C15
explain: plate + time + topology

Edge-first pipeline

Video stays local. Upstream is structured JSON only.

Edge→Graph→Console

Unified traffic analytics

Vehicle journeys and corridor / OD metrics from one event stream.

Corridor travel time +2.1 min
OD matrix updated · hour 09
Camera C07 health OK

Architecture

From camera feed to city-wide decision.

01

Ingest

RTSP / ONVIF · existing CCTV & ANPR

02

Edge

Detect · track · OCR · attributes

03

Event

JSON schema · no raw video upstream

04

Graph

4-signal fusion · spatial-temporal links

05

Console

Map · search · alerts · analytics

Latency

Edge-first · sub-second local inference

Privacy

Video on-prem · events only upstream

Scale

Corridor → district → city rollout

Use cases

“Show me this plate since 09:00 across the corridor, and tell me why you matched C07.”

Traffic operator

“Give me travel time on C03→C15 and the OD for this hour, from the same events.”

Mobility analyst

Principles

How NEXUS is built to operate.

I

Events, not video

Upstream traffic is structured JSON. Raw feeds never leave the edge node unless policy explicitly allows it.

II

Never force identity

Every cross-camera link is scored and labeled: MATCH, PROBABLE, UNRESOLVED, CONFLICT. Operators see why.

III

One graph, two products

Vehicle journeys and city traffic metrics derive from the same spatial-temporal graph. No duplicate pipelines.

IV

Human before action

Consequential alerts require operator acknowledgment. Automation assists; humans decide.

FAQ

Frequently asked questions

No. NEXUS is a layer over CCTV and ANPR you already own. Each feed becomes an edge node that emits events, not video.