Deep learning video analytics for the assessment of street experiments: The case of Bologna

Abstract

Planning infrastructures and services for sustainable urban mobility is one of the main challenges for European cities. Following these principles, in March 2022, Omitted for blind peer review and the Omitted for blind peer review built a new public space for children in a dismissed parking area located near to a middle school, using the approach of tactical urban planning and participatory design. In this context, this study proposes a methodology for the integration of long term camera-based monitoring for the assessment of temporary streets experiments. The method differs from previous work from literature in its continuous and systematic approach, fostering the implementation of large scale quantitative methodologies in urban interventions. The area was the subject of an extended mobility study with the objective to monitor pedestrian and vehicular flows through video analytics techniques and to detail specific patterns of space use during the pre/ post-intervention phases (data collected over two months). The results of the analyses were processed in order to: (i) identify the activation times of the study area, through metrics describing cumulative pedestrian density and cumulative dwell time; (ii) characterize the relative uses of the study areas following the plaza redevelopment. Results revealed a 43 % growth in the cumulative dwell time recorded in the area following the redevelopment intervention. New use characteristics related to furniture in the plaza emerged, with the most significant cumulative dwell times increase being recorded around these spots. The results presented in this research have made it possible to quantify the effectiveness of the proposed urban regeneration intervention, confirming the use of sensors and innovative analysis technologies to support the iterative design process of urban regeneration interventions.