Formula 1 Grand Prix crowd monitoring with Fyma

Operational capacity planning is something that during large events can be an absolute nightmare – from getting accurate, real-time data across complex urban terrain to then responding to it in a timely and proportional manner.

To this day many large events count their pedestrian traffic flow with old school clickers, the accuracy of which is questionable. We at Fyma decided to challenge this and test out our solution to provide more realistic actionable data. Because any camera in the world can be connected really quickly, we collaborated with Dutch company NS Stations in doing just this – helping them using drones to get real-time insights on crowds moving between the F1 grounds and their stations to inform capacity management for trains.

Crowd monitoring with drones and Fyma AI

The Formula 1 event took place on the Zandvoort circuit from 3 to 5 September 2021, with a capacity of 67% compared to full occupancy, good for approximately 70,000 visitors per day.

To ensure that the flow of these groups runs smoothly and that public transport is not overloaded, an innovative pilot was carried out using a drone.

The drone itself did not have any way to stream the data to Fyma so a laptop was used to setup a RTMP server and stream the data to Fyma.

Drones landed on and off the roof terraces of the NS Station Zandvoort aan Zee

The aim of the pilot was to map the traffic flows from the circuit to one of the train stations leading away from the circuit using Fyma. By applying Fyma’s AI to the images from the drone, it was investigated whether it is possible to make a better estimate of the flow and safety of large groups of people. It also looks at the results at different heights, camera types and perspectives.

We investigate whether the algorithm is able to count people from a crowd. By setting up different measurement zones, we hope to measure not only the number of people, but also the speed of the crowd in certain zones. Based on this information, we can dose the flows better and ensure that the station and the trains do not become too crowded. Because the AI processes the images in real-time, the images do not need to be saved. This also makes this solution privacy-friendly.
Jeroen van den Heuvel, Project Leader

What equipment was used?

A DJI M210 V2 with the DJI Zenmuse X7 camera was used for this operation. Due to the fact that the drone had to stay in a fixed position for 2 hours, an Elistair Safe-T 2 cable system was chosen, which removes the limitations of the flight time (approx. 25 minutes). The Elistair Safe-T 2 is equipped with a 100 meter cable, an auto-retract system and various safety functions such as an app with which important telemetry can be read.

Monitoring traffic in 20 seconds

To guarantee the safety of visitors and to keep the impact of the operation as low as possible, a take-off and the landing area was chosen on the roof terrace of the station building of NS Station Zandvoort aan Zee. This allowed a physical separation of the operational area and the pedestrian area to be ensured by the roof over the pedestrian area of the station. This part of the station was closed to the public, but it also protected the safety of personnel and security. 

The data processing was done by Fyma. We have developed advanced algorithms for monitoring traffic flows in urban environments. The images were sent to the server in real-time, analyzed by the algorithm, and sent directly to the NS in the Netherlands. This entire process takes about 20 seconds.

Increased safety and ease of travel for train passengers

The first phases of the pilot have been successful. By using drones, an overview can be created immediately and the situation can be responded to more quickly. This allows NS Stations to better facilitate the flow of large crowds and increase the safety and ease of travel for train passengers.

The testing took place over an hour and the platform managed to detect around 4000 people split between six 5 minute flights.

During the test, the camera and drone detected the movement of 4000 people

Crowd monitoring without detecting human faces

Fyma has been built in a way that does not allow for any kind of biometric data gathering whatsoever. The platform ensures privacy on multiple levels: the algorithms have not been trained to detect human faces and have never seen one – they simply look at a silhouette from the neck down; in a flying drone scenario the faces of people are not distinguishable and the AI sees people as a flock rather than individuals. 

The platform is GDPR compliant and an individual’s consent is not needed for analytics gathering as no biometric data is gathered. Fyma does not store any video data – only metadata is kept. Processing takes place in our secure servers in Frankfurt and never leaves the EU.

fyma detection brackets

Ready to unlock the data
from your existing video feed?

monitorBook a demo today
fyma logo
Contact us at
For press enquiries:

By using this website, you agree to our use of cookies. We use cookies to provide you with a great experience and to help our website run effectively.