HUMAN OBSERVATIONS NO MORE: MEASURING PLAYGROUND USE WITH COMPUTER VISION

Background: Observations of children’s play are crucial for evaluating the use of playgrounds. Today, human-based direct observation tools such as the System for Observing Play and Leisure Activity in Youth (SOPLAY) are commonly used. These tools have limitations: they are resource-intensive, the scalability is limited, they do not allow long-term measurement, and there is a risk of human-induced bias. Aim: In theory, cameras-based observations and computer vision can effectively address these challenges. However, in practice there are still many barriers. We are developing a solution resolving the quirks and obstacles we have encountered. Methods and Results: Preliminary analysis is underway, utilising footage from outdoor playgrounds by a video camera processed through the YOLO computer vision model. Validation is achieved by comparing the model's automated detection of time spent and occupancy in specific areas of interest (e.g., equipment) with human-annotated ground truth data. Conclusions: We have successfully demonstrated our concept's feasibility and identified potential users eager to adopt our solution. However, to meet the specific needs of both our target users and the research community, further development, validation, and rigorous real-world testing are required. To accelerate this process, we are actively seeking collaborative opportunities.