Examining the effects of intersection attributes on the rate of speeding events under daytime and nighttime conditions- A video-analytics study

Author(s): Jafari Anarkooli, Samara, Miranda-Moreno, St-Aubin

Slidedeck Presentation:

Slide deck link



Speeding is widely accepted as one of the most dominant factors determining the likelihood and severity of collisions. Among others, time of day is a factor that may affect the driver’s choice of travel speed. Several studies have explored the effects of time of the day on speeding behavior, mainly using independent variables representing different time intervals, and have consistently shown that the speed limit violation frequency is disproportionate during different times of the day.


To shed additional light on speeding behavior, we have previously examined how the rate of speeding at signalized intersections changes during different time intervals. Results show that drivers are more likely to exceed posted speed limits under nighttime conditions, as opposed to daylight conditions. This study builds upon our previous research by aiming to uncover different intersection attributes affecting the rate of speeding violations under daytime and nighttime conditions, separately.


The TrafxSAFE software has been employed to extract the speeding data. The software uses computer vision and artificial intelligence to detect and classify road users up to 13 categories (passenger, pick-up truck, pedestrian, etc.). These road user detections are then tracked across multiple frames to provide road user trajectories. The data used in this study includes speeding information for over 32 million road user trajectories from 58 signalized intersections for 8 cities in North America. Separate models for daytime and nighttime conditions were then developed to establish relationships between the speeding violations and the intersection attributes.


The preliminary modelling results indicate that traffic flow is significantly correlated with the rate of speeding for both daytime and nighttime condition models. As is expected, drivers were found to be more likely to violate speed limits at lower volumes, either in daylight condition or at night. Moreover, the results show that under conditions where traffic volumes are corelated with low built environment densities, speeding for both daytime and nighttime conditions is reduced. Interestingly, the results also suggest that while posted speed limit is positively associated with the increased rate of speeding, it is only statistically significant in daylight conditions.


The factors contributing to the increased rate of speeding were found to be generally consistent during nighttime and daytime conditions. While these results do highlight the importance of attributes such as traffic volume, further analysis is needed to uncover the complex interactions of different factors with the likelihood of speeding in different times of the day.


This research is mainly targeted at improving road safety for all road users through providing insights on speeding behavior of drivers under different time of day. Policy makers can intervene by, for example, adjusting the posted speed limits for intersections that meet certain critical criteria that encourage drivers to travel at higher speeds. As another example, traffic engineers may take advantage of the results to implement smart road lighting that can be used to attenuate speeding during nighttime conditions at signalized intersections.