How vibrant is this space? Identifying pedestrian social interactions in urban space through computer vision

As rapid urbanization transforms cities, creating public spaces that enhance social interactions and economic vitality has become a priority. Traditional urban vitality assessments, focused on pedestrian counts and static usage, often overlook the social interactions that contribute to economic activity. This study addresses this gap by developing an optimized social interaction recognition model based on YOLOv7, integrating interaction duration, social distance, movement status, and participant count from CCTV footage. To better link urban design with regional economic development, we introduce the Social Interaction Intensity Index (SIII), combining interaction frequency and duration to assess social engagement. By identifying high-interaction zones, policymakers can strategically allocate Levelling Up Funding to enhance public spaces that foster business activity, tourism, and local commerce. This study aligns with evidence-based urban planning, demonstrating how computer vision and spatial analysis can guide economic regeneration. While limitations such as reliance on pre-recorded footage exist, a robust multi-period data framework mitigates these challenges. Our findings support a targeted approach to urban investment, ensuring that funding is directed towards areas where improved public spaces can drive economic growth. By uncovering the relationship between social interaction and economic vitality, this research provides a data-driven foundation for equitable, sustainable regional development.