People counting by Hybrid Fusion

A people counting system measures visitor traffic (footfall) and behavior in a physical venue. It helps teams monitor entries and exits, understand live occupancy, analyze dwell time, and connect traffic to operational and business KPIs. Ariadne's Hybrid Fusion people counting does this without cameras: a patented signal sensor handles multi-zone tracking, while Time-of-Flight units validate counts at entrances, and the data feeds the same dashboard used in retail, airports, malls, and smart cities.

Footfall counting refers to visitor entries and exits over time. Occupancy is how many people are in an area at a given moment. Dwell time measures how long visitors stay in a zone, useful for understanding engagement and congestion. A modern people counting system measures all three from the same sensor mesh, plus the flow between zones, so retail, airport, and mall teams compare them on one timeline rather than reconciling three separate tools.

Ariadne combines a patented signal sensor with Time-of-Flight depth sensing. The signal layer carries continuity for multi-zone insights and journey analytics, since phones broadcast presence signals as they search for connections and the sensor reads those broadcasts anonymously. Time-of-Flight depth sensing supports device-independent counting at key points like entrances, where it gives the headline accuracy figure regardless of whether a visitor is carrying a phone. Fusing the two layers is what makes Hybrid Fusion accurate at the door and continuous across the zones.

Not necessarily. Some people counting systems use RGB video cameras, while others use privacy-first approaches like depth sensing, beams, or hybrid sensor fusion. Ariadne's people counting is camera-free by design: signal sensing detects radio emissions, Time-of-Flight measures distance, and neither method observes a face. There is no video footage to store, blur, or delete, which removes a significant GDPR and EU AI Act burden from the procurement file.

Ariadne People Counting delivers up to 99% accuracy in optimal conditions and 95%+ in typical retail and airport deployments. Accuracy depends on environment, density, sensor placement, and calibration; published competitor numbers typically sit between 85% and 99% with similar dependencies. A short pilot in a representative site validates the baseline before scale-up, and Hybrid Fusion holds its accuracy in low light and dense crowds where camera-based people counting tends to degrade.

Yes. People counting is most valuable when connected to operational and business systems. Ariadne supports dashboards, exports, and APIs to connect footfall, occupancy, and dwell-time data with BI tools, POS systems, scheduling tools, and reporting workflows. Out-of-the-box integrations cover the major retail and airport stacks; the API covers everything else, with native exports for teams that prefer a flat data path into a warehouse.

Ariadne's architecture sits structurally outside the EU AI Act's Annex III high-risk biometric categories. The patented phone signal sensor detects emissions, not biometric identifiers. Time-of-Flight measures distance, not appearance. MAC addresses and device IDs are not collected by default. Because no biometric data is captured or processed, the people counting system doesn't fall under biometric identification or biometric categorization classifications, and procurement teams in EU retail, transit, and public-sector deployments can evaluate Ariadne without the AI Act questionnaire friction that camera-based vendors face.

Yes. Camera-free people counters have shipped for over a decade, first with infrared beams and break-beams, then with Time-of-Flight depth sensors and Wi-Fi-based occupancy estimation. Ariadne's Hybrid Fusion is the next step: a patented phone signal sensor for multi-zone continuity, plus Time-of-Flight at entrances for absolute counts. Neither method captures an optical image. A camera is one of several ways to count; on a privacy and regulation file, it is the most expensive way.

Stereo-vision people counters claim 95 to 98 percent at the door under good light. Hybrid Fusion measures 98 to 99 percent at the door because Time-of-Flight is lighting-independent and signal sensing fills the dead zones between camera angles. Across a venue, the gap widens: stereo-vision loses accuracy at high ceilings, in dense crowds, and at the boundary between two cameras. Hybrid Fusion holds 95 percent plus in retail and airport conditions because the signal layer carries continuity where any single lens cannot.

It depends on the architecture. Camera-based people counting typically triggers a Data Protection Impact Assessment because live video is personal data under Article 4 of the GDPR, and large-scale monitoring of a public space sits in the Article 35 list. Camera-free people counting that never captures personal data, like Ariadne's Hybrid Fusion default deployment, does not meet the DPIA threshold on its own; ICO and CNIL guidance both note that genuinely anonymous footfall counting is a lower-risk operation. Customers should still document the assessment.

Regulation (EU) 2024/1689, the EU AI Act, classifies AI systems by risk. Annex III lists biometric identification and biometric categorization as high-risk, with Article 5 prohibiting certain biometric inferences entirely. Camera-based people counting that runs face detection or demographic inference is exposed to both. People counting that never captures a biometric identifier, like Ariadne's Hybrid Fusion architecture, sits outside Annex III. The Act treats the architecture, not the product label; a counter that observes radio emissions and distance is not a biometric system.

Yes, with calibration. Hybrid Fusion Time-of-Flight units distinguish stature bands, so a parent walking with a child holding hands registers as two entries by default, while a stroller registers as one entry plus a flagged event. Group counting rules are configurable per site: shopping centres often want strollers excluded from conversion-denominator counts, airports want every passenger counted whether they are adult or accompanied minor. Calibration during the pilot sets the rule; the dashboard exposes a per-zone toggle.

Time-of-Flight is lighting-independent because it measures distance, not appearance. Dawn, dusk, backlight, and sodium-vapour parking-lot light do not change the count. Camera-based people counters are sensitive to all four conditions: low contrast at dawn, shadow stretch at dusk, lens flare against west-facing windows, and colour cast under non-white lighting all cost the model accuracy. The Hybrid Fusion accuracy band of 95 percent plus holds across the daypart. This is why outdoor and atrium retail deployments score so differently between the two architectures.

Not on its own. iOS 14 and Android 11 randomized the MAC address in probe requests, so a single device now broadcasts dozens of unique identifiers per hour. The classic Wi-Fi probe-sniffing count is statistically unreliable at the zone level and falls below 70 percent accuracy in most retail and airport venues. It still has a role as one input among several: Ariadne's signal sensor combines anonymous radio detection with Time-of-Flight depth sensing in the Hybrid Fusion mesh, so the count holds even when MAC randomization wipes out the older Wi-Fi-only signal.

A single-site deployment runs four to six weeks end to end. The first week covers floor-plan review and remote sensor planning; weeks two and three cover sensor install and network provisioning; weeks three and four cover calibration against ground truth. Validation against manual counts closes the rollout. Multi-site programs scale the same checklist in parallel; an airport terminal with security, gate, and commercial zones runs the same playbook with more sensors and a longer calibration window.