People Counting

People counting is an AI video analytics capability that measures how many individuals pass through an entrance, occupy a zone, or traverse a corridor over time — without identifying any individual. It is the foundational metric for retail footfall, transit ridership, and public-space occupancy analytics.

How It Works

A people counting pipeline uses:

1. Overhead or angled camera — positioned to capture each person clearly as they enter or move through the zone.
2. Person detection — a neural network identifies every person per frame.
3. Tracking — a tracker links detections across frames to avoid double-counting.
4. Line crossing logic — counts increment when a tracked person crosses a defined line.
5. Aggregation — counts are summed by hour, day, entrance, and direction.

Why It Matters

Footfall is the top-of-funnel metric for retail and public spaces — without it, conversion rates, staffing decisions, and real-estate ROI are guesses. AI people counting:

• Works on existing cameras — no dedicated counting sensors.
• Delivers direction data — in, out, and net occupancy.
• Supports occupancy caps — COVID-era and capacity compliance.
• Scales across sites — consistent methodology per store or station.

IncoreSoft's Heat Map module includes people counting as a core capability, feeding retail analytics deployments worldwide.

Use Cases

• Retail chains — footfall-to-sales conversion by store and hour
• Shopping malls — tenant attraction and common-area flow
• Transit hubs — platform and concourse ridership
• Event venues — capacity management and evacuation planning
• Office buildings — occupancy-based HVAC and cleaning

Frequently Asked Questions

How accurate is people counting?

Top systems achieve 95–99% accuracy at properly placed overhead cameras. Accuracy decreases with heavy crowds (occlusion), extreme angles, or poor lighting.

Is it privacy-friendly?

Yes — standard people counting uses anonymous detection, not face recognition. Only aggregate counts are stored, which is typically outside GDPR's biometric-data definition.

Can it count children vs. adults?

Basic people counting doesn't distinguish age. Combined with age and gender detection (where legally allowed), you can segment footfall by demographic.