Project 04 / 04 — In Development

Gym Occupancy Solution

Built for a fitness facility in New Zealand

A hybrid IoT occupancy tracking system that combines LoRaWAN door sensors with edge AI camera detection to give fitness facilities accurate, real-time room occupancy data. LoRaWAN sensors track entries and exits continuously, while an on-device camera provides independent headcount snapshots to catch and correct sensor drift.

1-min Data Resolution
24/7 Automated Monitoring
Gym occupancy dashboard showing real-time room occupancy chart with 5-minute resolution data
The Challenge

Manual headcounts don't scale

Fitness facilities need to know how many people are in each room at any given time. This matters for health and safety compliance, class capacity management, and understanding usage patterns. Traditional approaches rely on staff doing manual headcounts — which is time-consuming, inconsistent, and doesn't produce usable historical data.

Off-the-shelf people counters exist, but they come with limitations. LoRaWAN door sensors count entries and exits but can drift over time as small counting errors accumulate. A single missed count throws off the running total for the rest of the day. Facilities need a system that not only counts but also self-corrects — and one that gives staff the tools to understand and validate the data they're seeing.

the-problem.log
09:00 — Actual 12 people
09:00 — Sensor 12 people
14:00 — Actual 8 people
14:00 — Sensor 14 people (drifted)
The Solution

Hybrid counting with self-correction

Exact IOT built this as a hybrid counting system. LoRaWAN sensors installed at doorways track entries and exits in real time, while an edge camera running on-device object detection provides an independent absolute headcount. By combining these two methods, the system can cross-reference relative counts against ground truth snapshots, catching drift before it becomes a problem.

All data flows into a web dashboard purpose-built for facility staff. The dashboard shows occupancy charts at 5-minute resolution, highlights peak periods, and supports day-by-day navigation. Staff can also log manual counts for validation, and the system includes a class scheduling tool so occupancy data can be viewed alongside the studio timetable.

Dashboard occupancy chart showing real-time room headcount data

Under the hood

The system combines LoRaWAN people counting sensors with a Raspberry Pi running on-device object detection. All camera inference runs locally — no video or images leave the premises. Data flows through Azure Service Bus into Azure Functions for event-driven processing, with results stored in SQL Server and rendered on a responsive web dashboard. Currently deployed at a single pilot site with plans to expand to additional locations.

C# .NET Azure Functions Azure Service Bus LoRaWAN Raspberry Pi Edge AI Chart.js SQL Server
occupancy.specs
Door Sensors LoRaWAN counters
Camera Hardware Raspberry Pi + camera
Edge AI On-device, no cloud
Connectivity LoRaWAN + WiFi
Cloud Platform Azure
Data Pipeline Event-driven (Service Bus)
Dashboard Web app + Chart.js
Resolution 1-minute intervals

Need accurate occupancy data
for your facility?

We build custom IoT solutions that combine sensors, edge AI, and cloud dashboards — tailored to your space.