Finland examines wider use of AI video camera technology

Finland is examining the wider deployment of artificial intelligence (AI) technology to support crowd control, immigration resource analysis and forward-looking regional population planning.

The country’s Interior and Justice ministries have established a joint expert group to examine the legal ramifications of using AI to monitor large gatherings and “individual movements” in outdoor spaces in urban and rural settings.

Finnish municipalities and police authorities are interested in deploying AI systems to deliver more precise “crowd headcounts” in urban settings, especially as part of the security infrastructure around large public assemblies such as music festivals and sports events.

The City of Helsinki (CoH) became Finland’s first urban district to launch an AI-enhanced camera “crowd counting” project in August 2020. The AI video camera surveillance technology was pilot-tested in Helsinki during August and September in partnership with smart tech company Viria.

Two popular city centre sites were chosen for the test, including the Senate Square in Helsinki’s oldest quarter. In particular, the CoH is examining the potential to deploy AI video camera technology to monitor large crowds with capacity thresholds, such as music concerts and sports events.

As public safety measures, the pilot-tests will provide the CoH with practical information about the usability and functionality of AI video camera systems in tracking crowd numbers. In this scenario, the technology would be deployed to alert police and rescue authorities should numbers breach set limits. It would also enable event organisers to intervene before public spaces become congested.

The CoH moved swiftly to allay concerns raised by Finnish human rights groups that the AI camera surveillance test system might violate citizens’ privacy rights.  

The authority issued a public assurance that the system being tested was only permitted to collect human images as data blocks rather than images of individuals.

Also, the technology used algorithms to calculate crowd sizes based on sample images. It was equipped with a wide-angle camera capable of taking a continuous video image of the defined site area.

The CoH used the AI video camera test system to support the Finnish capital’s Covid-19 containment operations during August and September. The technology helped its efforts to maintain close tracking of congestion in public spaces, said Anssi Vuosalmi, chief of contingency operations for the City of Helsinki.  

“The two sites where the AI solution was tested had caps on assemblies of over 500 people,” he said. “The numbers of people monitored during the trials ranged from one to 500. In most instances, the gatherings involved 12 to 200 people. In terms of privacy, camera images were not viewed by anyone other than machine intelligence processed by algorithms. We only saw numerical values rounded off to the nearest tens.”

The pilot-test is part of the CoH’s long-term plan to broader the deployment of AI-linked video analytics and surveillance technology to measure gatherings in public spaces more accurately. The authority hopes to partner law enforcement and emergency rescue service organisations to develop an integrated AI video camera surveillance capacity to police large gatherings.

In the Helsinki trials, crowd sizes were measured at 10-minute intervals, with the information, including images, stored in a central database, said Mika Rissanen, Viria’s business development director.

“The video cameras and the associated AI measured the average number of people in the field of view over a 10-minute period. Following the measurement, the personal data was stored in a database and a real-time visual snapshot was produced from the data,” he said.

The use of AI-based technology for crowd monitoring will become more widely used and more accepted, said Timo Heikkinen, CEO of Helsinki-based Top Data Science (TDS).

“The demand for video-enabled AI technologies that can provide accurate images, calculate human numbers and also protect personal privacy will certainly expand in coming years,” he said.

TDS recently closed a deal to deliver a privacy-preserving AI video analytics system to a property management company in Japan. The technology has the capacity to monitor spaces, count visitors and calculate their distances in indoor and outdoor locations.

The system, which can be modified to enable mask detection, uses novel image-to-image learning and 2D density probability predictions to ensure privacy without identifying individuals.

TDS developed the technology primarily to provide reliable density information in real time at locations such as shopping malls, restaurants, offices and other public places based on high-definition camera images. It is also designed to send alerts when the number of visitors in a specific area has exceeded a certain threshold.

TDS’s contract reflects the increased demand for AI video camera technology as organisations implement strategies to improve social distancing awareness and conform to Covid-related public health rules without compromising privacy rights.

Meanwhile, next-generation AI systems are being rolled out to support the operations of Finland’s Digital and Population Data Services Agency, which is using advanced AI and digital technologies to help manage changes in demographics and population migration.

The agency’s improved AI digital systems are being used as micro-planning tools to track population changes across Finland’s cities and provinces more precisely. The aim is to use AI to help match state investment and spending with demand for public services.