School zone safety diagnosis using automated conflicts analysis technique
School safety is a high priority for road safety agencies worldwide due to the growing concerns about student safety and emphasis on community livability. However, the lack of consistent data available at schools makes it challenging for engineers to understand the safety issues faced accurately. Traffic conflicts have been advocated in the literature as a surrogate safety measure due to the advantages it offers for road safety evaluations. This paper demonstrates the capability of automated traffic safety diagnosis at a school using computer vision techniques. The selected school is located in a residential area in the City of Edmonton, Alberta, and is on a main roadway in the neighbourhood. The age group of the school is from 5 to 16 years old (corresponding to Grade 1 to 9 respectively). Data are collected during the fall and winter terms. The severity and frequency of conflicts and traffic violations were analyzed to quantify the safety concerns. These concerns included driving violations, jaywalking violations as well as conflicts between pedestrians and vehicles as well as between vehicles and each other. Hourly and seasonal trends were observed and analyzed to assist in the selection of treatments and recommendations to improve the safety around schools. The results show that pedestrian safety has improved in winter due to lower vehicle speeds.