Pedestrian safety remains a major challenge in densely populated urban areas and in touristic regions, where traffic dynamics and human behavior vary significantly. In Greece, both the center of Athens and popular tourist destinations across the country experience increased risks for vulnerable road users, especially at signalized intersections. This study investigates pedestrian non-compliance at two locations with contrasting urban characteristics. The one is a signalized intersection near Syntagma Square in the center of Athens, and one in Rethymno, Crete, a highly touristic city. Data collection was performed using video recordings from smartphone cameras, demonstrating a low-cost and flexible approach for urban traffic monitoring. Video data were processed through an advanced computer vision algorithm, structured in the context of the Phoebe Project, integrating YOLOv8, ResNet-50, homography transformation, and Kalman filtering, obtaining precise tracking of pedestrian and vehicle trajectories. Comparative analysis between the two locations illustrates the differences in non-compliant pedestrian behavior induced by the effect of urban density and tourism-related factors. The conclusions of this study enlighten pedestrian safety concerns in several Greek urban spaces and reveal the potential of video recognition algorithms to aid road safety research.
| ID | pc655 |
| Presentation | |
| Full Text | |
| Tags |
