Video-based Pedestrian and Vehicle Traffic Analysis During Football Games

TL;DR

The study develops a video-analytics pipeline that uses YOLOv4 for detection, DeepSORT for tracking, and fisheye-to-rectilinear transformations to quantify pedestrian and vehicle trajectories at a UF campus intersection during football gamedays. It computes Time-To-Collision ($TTC$) and Post-Encroachment Time ($PET$) to characterize pedestrian–vehicle and vehicle–vehicle conflicts, and classifies P2V conflicts into six types, enabling spatial-temporal analysis via heatmaps and KDE. The results show gamedays increase both pedestrian and vehicle volumes, with P2V conflicts peaking several hours before kickoff and V2V conflicts not rising correspondingly; win-probability data from the away team correlates strongly with volumes ($r$ up to $0.93$, $p$ < $0.02$), suggesting predictive signals for traffic management. The work proposes countermeasures such as a Barnes Dance phase and enhanced enforcement to improve safety on event days, illustrating how event-driven traffic analysis can inform proactive intersection control.

Abstract

This paper utilizes video analytics to study pedestrian and vehicle traffic behavior, focusing on analyzing traffic patterns during football gamedays. The University of Florida (UF) hosts six to seven home football games on Saturdays during the college football season, attracting significant pedestrian activity. Through video analytics, this study provides valuable insights into the impact of these events on traffic volumes and safety at intersections. Comparing pedestrian and vehicle activities on gamedays versus non-gamedays reveals differing patterns. For example, pedestrian volume substantially increases during gamedays, which is positively correlated with the probability of the away team winning. This correlation is likely because fans of the home team enjoy watching difficult games. Win probabilities as an early predictor of pedestrian volumes at intersections can be a tool to help traffic professionals anticipate traffic management needs. Pedestrian-to-vehicle (P2V) conflicts notably increase on gamedays, particularly a few hours before games start. Addressing this, a "Barnes Dance" movement phase within the intersection is recommended. Law enforcement presence during high-activity gamedays can help ensure pedestrian compliance and enhance safety. In contrast, we identified that vehicle-to-vehicle (V2V) conflicts generally do not increase on gamedays and may even decrease due to heightened driver caution.

Figures (25)

• Figure 1: The pipeline and deep model architecture used in this paper that has been adapted from our previous work huang.
• Figure 2: Case 1 in TTC computation involves scenarios where either one or both objects are stationary, or one of them is in motion. When at least one object is moving, a potential conflict arises only if the path of movement aligns with the line connecting the two objects.
• Figure 3: Case 2 of TTC computation involves scenarios where both objects are moving along parallel lines. In this case, a collision is only possible when the lines are coincident, meaning they are directly on the same path, in which case TTC is calculated using the formula: TTC = distance between the objects / relative velocity of the objects.
• Figure 4: Case 3 is the most general case of conflict occurrence, which involves two objects approaching each other from different directions. The conflict point, where their lines of motion intersect, is first determined. The distance of the objects 1 and 2 from the conflict point are represented as s1 and s2, respectively.
• Figure 5: Overview of PET computation process, involving mesh division of the intersection region, and object list creation. PET events are identified when objects interact and move over the same region within a short period, potentially indicating significant events.