Demographic Detection
Description
This metric classifies the visitors crossing an entrance into demographic categories — man, woman, child — producing an anonymous, aggregated profile of who visits the space, by day and hour.
| Element | Detail |
|---|---|
| Description | Classifies each person counted at an entrance (People Counting) into Man / Woman / Child, and aggregates the results per time bucket. |
| Algorithm | 1. Detection: AI anonymizes and detects people crossing the counting line. 2. Classification: a neural network assigns the demographic category from abstract visual features. 3. Aggregation: only counters per category and time bucket are stored — never individual records. |
| Categories | • Man • Woman • Child |
| Key Parameters | • Enabled per counting line (typically entrances) • Time bucket for aggregation (hour, day) |
| Privacy | No facial recognition and no biometric identification: the classification runs on anonymized data in real time, the source images are discarded immediately, and only aggregated counters are stored. See Privacy & Data Protection. |
| Typical Use Cases | • Understanding the visitor profile per store, mall, or zone • Adapting assortment, campaigns, and windows to the real audience • Comparing the visitor profile across locations or time periods • Measuring how campaigns change the audience mix • Tenant mix decisions in malls |
Interpretation
- The output is a distribution (e.g., 46 % women, 38 % men, 16 % children) per time bucket — not individual events.
- Comparing the distribution across day/hour windows reveals audience patterns (e.g., families on weekend afternoons) that can drive staffing, campaigns, and store layout.
- Combined with Conversion Ratio Metrics, it shows whether the audience that visits is the audience that buys.
Related Metrics
- People Counting — the base counting on which classification runs
- Group Detection At The Entrance — complements the profile with group composition (individuals, couples, families)