What can AI-technology tell us about the use of neighborhood meeting places?

A monitoring system with embedded AI-technology might be a useful tool for municipalities to understand activity patterns at meeting places, which can inform future municipal planning and priorities.

Background: Many Norwegian municipalities invest resources into developing meeting places to promote activity without really knowing whether the places are being used by the residents. An important reason for this is that assessing use of meeting places is a challenging and time-consuming task for both municipal employees and researchers, especially if data is to be collected over a longer time period. Thus, we aimed to test an AI-based monitoring tool at a local meeting place to explore when and how much the place is being used and whether there are seasonal differences in usage.

Methods: A longitudinal monitoring study was conducted as part of the project Co-creating public health. In September 2023, a camera was installed at Reinsparken, a newly established pump track park and meeting place in Vinje municipality. The camera has a tiny built-in computer with an embedded AI-solution that allows us to quantify moving objects directly from a video image flow. The AI-solution reads approximately 30 images per second, and moving objects are continuously detected in each image. After detection is completed, video images are deleted every second and data is automatically transferred to Power BI that manage and analyze big data using AI-capabilities. In this way, no personally identifiable information is stored. Through Power BI, we have obtained data on monthly, daily, and hourly passings, as well as movement tracks. Up to date, we have collected and analyzed data over 98 consecutive days, from 2nd October 2023 to 7th January 2024. These data were linked to daily weather data from the Norwegian Centre for Climate Services. Descriptive statistics were computed, and heat maps were used to visualize movement patters at the meeting place.

Results: There were 804 daily passings at the meeting place over the monitoring period. The highest number of users was observed in week 46, with 2100 daily passings on average. Highest frequency of activity was observed in the month after the opening. In the December, we witnessed a substantial drop in daily passing (mean=93), compared to October (mean=1516) and November (mean=1102). To date, there has been most activity at the meeting place from Friday to Sunday, between 9 am and 1pm. Interestingly, we observed a change in the users’ movement pattern from week 46, when the closest weather station showed that the depth of snow had reached 15 cm.

Conclusion: Through the AI-based monitoring tool, we have captured when and how much the meeting place has been used over a 98-day period. Although a longer observation period is needed to fully understand user and movement patterns, our results indicate that this is a place people visit on weekends and that season characteristics influence how the place is being used.