COVID-19 Improve your event’s safety with AI
After 10 years working as Safety Manager at different festivals in France 🇫🇷 (>100 000 attendees) and 8 years in IT, I decided to work on a model to help events producers during the COVID-19 pandemic.
For many years, Safety Managers were trained on how to deal with virus exposure during an event (Measles, Mumps…) for which a vaccine is protecting most people in the crowd.
Today, we are facing a virus without vaccines and for which everyone is vulnerable. Government Safety Protocol is changing every day, and most countries around the world decided to cancel all the events above 5000 people.
Because we know that we cannot stop a crowd from dancing, here is how I manage an event during the Coronavirus pandemic.
Government and Regional Health Agency rules
In this article, we will focus on the latest French government and Regional Health Agency and the section that AI can solve:
- Force minimum 1 meter between people in the waiting line 📏
- Check the body temperature at the entrance 🌡️
- Check if the masks are wearing during all the event 😷
Main Entrance Management
In order to reduce the risk, you need to be sure that your crowd is not packed at the entrance. It is quite a challenge for events receiving more than 5000 attendees per day.
That is why we imagined to re-think our Main Entrance Management by copying the Plane Boarding Process ✈️ and including boarding section on the tickets. People are selecting the most suitable time slot for them to “board” the event.
By doing this, you avoid a massive crowd trying to enter right before your headliner and manage your communication before the event to ask people to be on time.
Temperature Body Screening
To screen the attendees’ body temperature at the entrance, you can use a fixed and portable thermal camera. Their usage is quite easy, and you can get a reliable metric (± 0.2°c).
A camera can scan up to 10 people at the same time in a range of 0.5/1.5 meters.
The best practice is to set-up a protocol on two levels depending on the body temperature level.
Level 1
The attendee can go ahead with the body check without double-checking.
Level 2
The attendee must go to a specific Covid-19 suspicion line managed by your safety provider (Red Cross, medics…) and not be allowed to enter the event.
Live AI models on your cameras
When your attendees are inside your event, you need to check that they are wearing a mask all the time.
However, it is impossible to ask someone not to drink, eat, and smoke during an event. Plus, most of the revenues are generated by Food Trucks and Bars.
In addition to your security guys patrols, you need to help yourself with an AI model on your CCTV on 3 levels.
Level 1
Attendee is wearing mask, no alert.
Green label: /
Action: Nothing to do
Level 2
Attendee made a movement to take off the mask (movement model + mask recognition model)
Orange circle: might be drinking/eating/smoking
Action: check later if it turned green
Level 3
Attendee does not have a mask
Red circle: no mask detected
Action: ask security patrol to go and ask to put the mask on
AI models are incredibly efficient on live footage, even if your screen is showing 10 000 attendees, the model will compute information way faster than a human brain.
For this AI Model, I am working with Picsellia. A French Startup specialized in Computer Vision.
Conclusion
AI can not solve all the problems linked to COVID-19, but at least, it can help. I hope this article helped you imagine your own Safety Protocol and helped people party without risks.
Let me know if you have other best practices for your events, sharing our knowledge is the best way to improve our Safety Protocol.
Thank you for taking the time to read my article 👋
