ConductVision Enterprise

Enterprise video analytics

ConductVision runs detection on the cameras you already have and reports the endpoints your team chooses, from PPE compliance to queue times, in a pilot on your own footage.

A busy shopping street in Nakuru, Kenya, with people walking between shops and parked cars; boxes mark 20 pedestrians.20 detections
Busy street in Nakuru
  • Person20

ConductVision detections on this photo. Photo: Victor Kibiwott, CC BY-SA 4.0, cropped.

How a pilot works

Endpoints first, then footage, then a decision

  1. Choose

    One outcome and its endpoints

    Pick one problem worth measuring, then agree the endpoints, cameras and zones before anything is installed.

  2. Measure

    A baseline, then your footage

    ConductVision records a baseline and runs on your cameras, and its detections are checked against a sample reviewed by hand.

  3. Decide

    A results review

    The review sets each endpoint against the baseline, and your team decides whether to scale, change or stop.

Capabilities

What the detection models review

Every image below shows what ConductVision's models found on that photo, at the settings recorded for it.

People and vehicles

Safety and security

Inspection and environment

Industries

Where visual monitoring fits your sector

Every pilot

The model is measured on your footage

Before anyone relies on ConductVision's numbers, each pilot reports how its detections compare with a sample your team or ours checks by hand. Detection runs on-premise or in your environment, and video does not leave your control.

  • Alert precision

    The share of alerts a reviewer confirms in a sample of alert clips.

  • Recall

    The share of events a reviewer finds in sampled footage that ConductVision also flagged.

  • False alerts

    Alerts a reviewer rejects, counted per camera per day.

Platform

Four layers from camera to action

  1. Ingest

    Live streams from RTSP and ONVIF cameras, recorded video and image uploads.

    • RTSP and ONVIF
    • Recorded video
    • Image uploads
  2. Process

    ConductVision's models for people, vehicles, PPE, and fire and smoke run on your hardware or in your environment, with models trained on your footage in a pilot.

    • On your hardware
    • Pilot-trained models
    • Zones per camera
  3. Analyze

    Detections become counts, rates, dwell times, heatmaps and trends, kept as an event log you can export.

    • Event log
    • Rates and trends
    • Heatmaps
  4. Act

    Alerts reach the person who can act, with the clip attached, and webhooks, REST and MQTT feed your other systems.

    • Alerts with clips
    • Webhooks and REST
    • MQTT

Pilot

Scope a monitoring pilot

Tell us about your operation, your cameras and the first outcome you want to measure, and a ConductVision engineer replies with a scoping plan.