Skip to content
Demo
  • Vision software

Next-generation tunnel AID

Automatic incident detection in tunnels on your existing cameras: stopped vehicle, wrong-way driver, pedestrian, debris, smoke - in a high-availability cluster, proven in production on the most demanding structures.

  • Real timedetection and alarm
  • Automaticcamera calibration and masks
  • 24/7high-availability cluster
Road tunnel: pedestrian detected on the emergency lane next to an intervention van

What Tunnelwatch does

  • Stopped vehicle and slowdown

    Immediate detection, including in dense traffic.

  • Wrong-way driver and congestion

    Immediate alert when the conditions are met.

  • Pedestrian, animal and cyclist

    Immediate detection and reporting.

  • Smoke and fire outbreak

    Early visual detection, ahead of physical sensors.

Incident detection in a tunnel: vehicles tracked by AI

Better accuracy for tunnel operators

Tunnels are subject to many constraints and traffic challenges; tunnel management is highly sensitive and monitoring can become costly, especially as safety equipment commands add up.

Thanks to the latest advances in AI, Tunnelwatch leverages intelligent video analytics to deliver robust incident detection that genuinely improves decision-making in control centers.

How it works

  1. Connect

    The streams of existing cameras are reused, IP or analog.

  2. Detect

    Analysis runs continuously on every stream.

  3. Report

    The alarm reaches the control room with visual verification, in your VMS or SCADA.

Active high availability

Unique on the market

Active-active cluster: instant takeover, zero monitoring interruption.

Unique on the market

Active high availability, not passive redundancy

A tunnel cannot go blind, not even for a few minutes. Cyclope is the only provider running AID in an active-active cluster: every node analyzes continuously, and the failure of one is absorbed without failover or monitoring gap.

The market standard

Passive redundancy

  • Active server
  • Cold standby
  • The standby sleeps: the failure must be detected, then switched over
  • Monitoring gap during failover - the tunnel is blind
  • The standby is only tested on the day of the failure

The Cyclope approach

Active-active cluster

  • Node 1
  • Node 2
  • Node 3
  • Every node analyzes continuously - no sleeping server
  • The load of a failed node is taken over instantly, with no failover
  • Zero interruption: every node is continuously proven in production

Accurate, open, easy to integrate

  • Automatic mask generation

    Automatic lane segmentation, at installation and over time, for adaptive maintenance of camera mask configuration.

  • Qualified data

    The AI recognizes object shapes, correlations and incident patterns, taking the overall context into account: space, time and simultaneity.

  • Cross-camera intelligence

    Thanks to object tracking, each vehicle is uniquely identified from one camera to the next, for double-checked alarms and better prioritization.

  • Dedicated interface

    A modern, easy-to-use interface for supervisors and integrators, to qualify the system and monitor more efficiently.

  • Open, equipment-agnostic system

    Compatible with most cameras and VMS on the market, with no specific equipment constraint - easier maintenance, unlimited evolution.

  • Gated deployment, shared O&M

    Factory acceptance, pilot site, field rollout - stable and versioned APIs, 24/7 monitoring, shared operator / maintainer dashboard.

Who is it for?

  • Tunnel operators

    A modern, intuitive interface that is easy to learn. By reporting only relevant and accurate data, it provides real decision support.

  • Infrastructure operators

    By efficiently filtering false alarms, our AI-powered AID improves safety in road tunnels, with significant savings in time and money.

  • Integrators

    A smart segmentation algorithm automatically generates masks tailored to each camera: Tunnelwatch simplifies system installation and maintenance.

Tunnelwatch interface: incident list, filters and incident card

High performance

AI at the service of performance

Building on cutting-edge expertise in the latest deep learning technologies, our detection algorithms have been trained on millions of images to recognize specific shapes and situation patterns: tracking, behavioral analysis and object detection.

Tunnelwatch interface: TMD passages and detection cards

New feature

Hazardous goods detection

Our video analytics AI detects the placards of vehicles carrying dangerous goods from a single camera. It reliably tracks the number of hazardous goods trucks on the network - a plug-and-play solution for better monitoring of these vehicles that require special attention.

Already in use

  • VINCI Autoroutes operator in front of the Duplex A86 Security Control Centre video wall
    Duplex A86 Security Control Centre
  • Operators in front of the Escota Security Control Centre video wall
    Escota Security Control Centre

Start a Tunnelwatch pilot

A defined scope, written performance criteria, an exit clause with no residual cost. That is how we start every project.