How Mansfield Used People Counting to Track Foot Traffic
The Hong Kong Tourism Board hosted a spectacular dragon boat event from June 27 to 28, drawing thousands of the world’s top dragon boat athletes to compete beneath the iconic Victoria Harbour skyline. While the event generated tremendous excitement and drew large crowds, it also presented significant operational challenges. Organizers needed accurate, real-time data on when and how people entered and exited the venue, as well as detailed peak-time distribution insights—all to support on-site crowd management and flow coordination.
This is where Mansfield Consulting stepped in. As the People Counting solution provider for the event, we deployed advanced computer vision technology to translate camera footage—people moving through the venue—into structured, actionable data: who entered, who exited, and exactly when those movements occurred.
Our People Counting Method: Detection + Tracking + Entry/Exit Counting Line + Time Aggregation
Our measurement workflow is built on a four-step process designed to ensure clarity, traceability, and seamless alignment with on-site operations.
Step 1: Person Detection
For every video frame captured by the cameras, the system detects individuals and outputs a bounding box for each person. This forms the foundation for all subsequent tracking and counting.
Step 2: Multi-Object Tracking
The system then links the same person across consecutive frames, assigning a unique track ID to each individual. This step serves two critical purposes:
Prevents double counting – avoiding errors caused by frame-to-frame changes or momentary occlusions
Reduces miss rates – ensuring that brief obstructions do not result in lost tracking
Step 3: Counting Rule Using a Measurement Line
At each entrance or passage area, a virtual measurement line (one line or two lines, depending on the specific setup) is defined. The system determines movement direction based on whether the tracked person’s center point crosses the line:
Crossing from one side to the other → counted as “Entry”
Crossing in the opposite direction → counted as “Exit”
This rule-based approach ensures accurate, consistent directional counting across all venue access points.
Step 4: Time Binning
Finally, all counting events are aggregated into fixed time intervals (e.g., every 15 minutes) using their timestamps. This produces clear time-segment statistics, allowing organizers to:
Observe crowd flow patterns with precision
Identify peak periods for proactive crowd management
Allocate staffing and resources more effectively
Turning On-Site Crowds into Useful Management Data
In large-scale events, effective crowd management requires more than just a total headcount. Organizers also need timing insights—such as when entry spikes occur, how exit patterns evolve throughout the day, and what the busiest windows look like. Mansfield’s measurement approach is specifically designed around these operational needs, transforming raw camera data into decision-ready intelligence.
Event Recap: Supporting the Hong Kong Tourism Board with Data-Driven Insights
Looking back at June 27–28, Mansfield is proud to have supported the Hong Kong Tourism Board in deploying a People Counting solution that operationalizes entry and exit measurement. By combining detection, tracking, entry/exit line counting, and time aggregation, we helped turn camera observations into clear, quantifiable, and actionable results.
The data we delivered enabled the organizers to:
Optimize real-time crowd flow management
Improve safety and security coordination
Enhance the overall participant and spectator experience
Let’s Discuss Your Event’s People Counting Needs
If you’d like to understand how the solution was optimized for this specific venue—including entrance placement, passage design, and camera angle considerations—we’d be happy to share more details.
Contact Mansfield Consulting today to learn how our computer vision-powered people counting solutions can support your next large-scale event.
