Data Processing

How Video Becomes Analytics

Digeiz transforms video streams into aggregated insights through a processing chain designed for privacy and operational reliability.

The platform connects to approved video sources, processes them locally in real time, extracts movement and behavioral signals, and delivers aggregated analytics through dashboards and APIs.

Cameras / VMS
Local processing
Technical transformation
Aggregation
Dashboards / APIs
1

Connect to approved video sources

The platform connects only to the video streams selected for analytics use. Depending on the deployment, these can come from existing security cameras or from cameras dedicated to audience measurement.

2

Process locally, in real time

Video is processed on image processing servers deployed on the client site. This keeps the most sensitive stage of processing close to the source and avoids unnecessary circulation of personal data.

3

Turn images into movement intelligence

The AI pipeline detects people, reconstructs trajectories, and extracts the signals needed to understand traffic, movement, dwell time and visitor pathways. For cross-camera journey analysis, the platform relies on appearance-based signals rather than facial identity.

4

Aggregate before delivery

The outputs delivered to clients are aggregated analytics used for reporting, optimization and integration. These include counting, segmentation, dwell time, typical pathways, cross-visits and media performance indicators.

What stays — and what does not

Used to generate analytics
  • Counting events
  • Traffic and flow indicators
  • Dwell time metrics
  • Pathway patterns
  • Aggregated audience insights
Not the purpose of the platform
  • Facial recognition
  • Named individual identification
  • Identity authentication
  • Long-term storage of raw video as analytics output

A processing model built for both insight and restraint

This processing model allows Digeiz to produce high-value analytics for physical venues while keeping the handling of personal data focused, short-lived and proportionate.