Smart Tourism In the Age of the General Data Protection Regulation: Capturing Rich Visitor Flow Data Without Risking Privacy Invasion
Edinburgh welcomes more than four million visitors a year, so understanding tourist behaviours is vital to the development of this sector, which contributes £1.2 billion to the local economy and supports 300,000 jobs. There are many technologies that can monitor, record and analyse visitor movements but the collection of personalised data, without the agreement of individuals, is often prohibited under the UK’s General Data Protection Regulation (GDPR). So this was the challenge that researcher Ricardo de Azambuja set himself on his TRAIN@Ed research fellowship programme at the University of Edinburgh in 2021 – a programme open to only 25 people each year to help guide experienced researchers in research excellence and transferable skills.
As an engineer with a PhD in robotics and artificial intelligence, Ricardo was keen to develop cutting-edge technology to observe, analyse and to better manage tourist flows at visitor venues, while still remaining GDPR compliant.
Ricardo explained: “Traditionally, a researcher would go to a tourist venue and manually take notes to understand what the visitors enjoyed the most, where they spent time, etc, but in this project we wanted to automate the task and use ‘deep learning’ algorithms to give us more detailed information.”
Rather than using an off-the-shelf ‘smart camera’ to monitor visitor flows where images are recorded on a hard disc or sent to the cloud, Ricardo’s solution was to create a device where the camera itself processed the information without ever saving or transmitting personal data; he called it the Maple Syrup Pi Camera.
He used a Raspberry Pi Zero W and a camera supplemented with the neural network accelerator Google Coral Edge TPU, which is capable of running complex computer vision-related algorithms to identify human movement or objects but that could run at very low power, less than 2W.
He explained: “A low-power device was essential as this would mean that the camera could be placed anywhere and run for at least a day, or longer with solar power. I chose the Raspberry Pi as it is one of the best-selling general-purpose computers on the market and therefore it has a huge community who can give advice and support to help develop applications.”
His smart camera undertakes all the data processing itself, online, so no data, such as personal images or vehicle number plates, is collected on a hard drive for processing later on; what is recorded is anonymised data used to create ‘heat maps’ of people movements and behaviours at various tourist venues.
While development of the device was successful, the commercial application at Edinburgh tourist venues was thwarted by the Covid pandemic, which coincided with Ricardo arriving from Canada in October 2020.
He said: “A number of tourist organisations were interested in trialling the device but the projects were put on hold because of Covid lockdown restrictions. The smart camera concept has great potential as it can be used for many applications. For example, we even developed a facemask recognition version to alert people entering a venue who were not wearing masks, and we had another project to use vehicle number plate recognition to map vehicle movements around a Scottish island to understand tourist behaviours.”
Ricardo had to return to Canada in November 2021 but, as his smart camera concept was developed on open source software, he’s hopeful that other researchers will develop the concept further.
For more information, visit https://github.com/ricardodeazambuja/Maple-Syrup-Pi-Camera
Read the latest Case studies
As one of the TRAIN@Ed programme fellows, Sarah Galey had the opportunity to work with…
Childlight, the data institute based in the University of Edinburgh, is exploring how data can…