Automated computer vision video analysis techniques were used to analyze video data during the operation of New York City’s Summer Streets Program at a major signalized intersection. The main objectives of this study were to diagnose pedestrian and cyclist safety issues during the shared space operation and to demonstrate the feasibility of the automatic extraction of road user data (e.g., pedestrian, runner, rollerblader, or cyclist) required for microscopic behavior analysis. Road users’ speeds and pedestrian gait parameters (step frequency and step length) were automatically extracted and analyzed. Results show that pedestrian walking speed was highest during the Summer Streets operation (1.49 ± 0.54 m/s) because pedestrians had more street space to use and slowest during normal operations (1.30 ± 0.22 m/s). Bike speeds were low during the Summer Streets event (3.62 ± 0.97 m/s), likely because of interaction with pedestrians, but these speeds increased during normal traffic operations. Pedestrians and cyclists moving in groups tended to be slower and confirmed results found in previous studies. The safety analysis was conducted with traffic conflict techniques. It was observed that the lowest rate of conflicts between pedestrians and cyclists and between cyclists was found to be during Summer Streets operations. In addition, an analysis of spatial violations showed that some road users were not observing traffic rules in the transition period after Summer Streets ceased to operate.