Analysis and Automatic Detection of Pedestrians Running Red Light at Signal Intersection

This article delves into urban road signal intersections, employing a combination of field research and survey methods. It synergizes with the logic regression theory to distill the critical influencing factors that could trigger pedestrian red-light violations, further subjecting their impact to a fitting analysis. Building upon this foundation, it leverages the principles of supervised learning and technology, utilizing five classifiers: random forests, decision trees, multilayer perceptrons, logistic regression, and naive Bayes. This comprehensive approach is aimed at training a pedestrian red-light violation classifier, facilitating automated detection of such behavior at signal intersections.