In this study, we introduce an innovative methodology for the detection of helmet usage violations among motorcyclists, integrating the YOLOv8 object detection algorithm with deep convolutional generative adversarial networks (DCGANs). The objective of this research is to enhance the precision of existing helmet violation detection techniques, which are typically reliant on manual inspection and susceptible to inaccuracies. The proposed methodology involves model training on an extensive dataset comprising both authentic and synthetic images, and demonstrates high accuracy in identifying helmet violations, including scenarios with multiple riders. Data augmentation, in conjunction with synthetic images produced by DCGANs, is utilized to expand the training data volume, particularly focusing on imbalanced classes, thereby facilitating superior model generalization to real-world circumstances. The stand-alone YOLOv8 model exhibited an F1 score of 0.91 for all classes at a confidence level of 0.617, whereas the DCGANs + YOLOv8 model demonstrated an F1 score of 0.96 for all classes at a reduced confidence level of 0.334. These findings highlight the potential of DCGANs in enhancing the accuracy of helmet rule violation detection, thus fostering safer motorcycling practices.

Proactive and holistic safety management approaches should consider multi-modal crash risk. Cyclist crash risk should be prioritized given the high-severity of vehicle-cyclist crashes. Cyclist crash risk is difficult to quantify given the sparse nature of cyclist collisions and collisions in general. There is thus a need to develop a more proactive approach for multi-modal road-safety management by leveraging new technologies. This study proposes a conflict-based methodology to estimate cyclist crash risk using autonomous vehicle data, extrapolating from observed conflicts to real-time dynamic crash risk. Using 87 hours of data from an autonomous vehicle dataset from downtown Boston (nuPlan), traffic conflicts were identified. A Bayesian Hierarchical Extreme Value model was created representing driver and cyclist crash risk over short time intervals. This allows for identifying the real-time crash risk of various intersections and mid-blocks, enabling route-level safety metrics. The spatiotemporal characteristics of crash risk were examined in this study. Routes with cyclist facilities were found to be safer for cyclists, on average, than those with shared facilities. However, substantial fluctuations in crash risk were observed at different time intervals, with the shared facilities sometimes being safer than those with painted or buffered bicycle lanes. This highlights the need for real-time safety monitoring. At the user-level, a safest route application was also proposed, allowing for an impedance function to be developed based on real-time crash risk and the comparison of any number of nodes and links along a particular route.