This repository offers code to reuse methodology and repeat experiments in the study "A Unified Probabilistic Approach to Traffic Conflict Detection".

🗺 This is a service class application software that for the poor areas which have bad traffic safety,the crowd which have lower safety awareness and the people which go out to an unfamiliar place.

This is the source codes and modeling data for paper: LSTM + Transformer Real-Time Crash Risk Evaluation Using Traffic Flow and Risky Driving Behavior Data

This repository shares information and guidelines to use the reconstructed trajectories of naturalistic crashes and near-crashes in the SHRP2 NDS.

Real-Time Crash Identification using Connected Electric Vehicle Operation Data

A novel spatial Machine Learning (ML) framework that integrates multiple ML models—Random Forest, XGBoost, and LightGBM—with geographically weighted regression to account for spatial heterogeneity. The framework has been successfully applied to intersection crash frequency modeling, achieving excellent performance.

Vision-based adaptive crosswalk system that extends pedestrian green time using YOLO tracking, homography speed estimation, and safety logic to protect elderly and slow walkers in real time.

Using Predictive Analytics to Improve Traffic Safety, by Team Knight Rider of NYU Stern MSBA Class of 2019

This is the WinBUGS codes for AMAR paper: https://doi.org/10.1016/j.amar.2025.100387. It includes codes for running Grouped Random Parameters Poisson-Lindley model with spatial effects and other baseline models (e.g., Grouped Random Parameters Poisson-Lindley model).

YOLO^2 is an inference pipeline using two YOLO models in succession. It is designed for detecting motorcycles with the number of passengers, and how many of these are wearing helmets.

SAVeD (Social media-based ADAS-equipped Vehicle event Dataset) shaoyan.zhai@ucf.edu

For New-Taipei-City-Traffic-Safety-Hackathon

Codebase containing the YOLOv7 and GRU model implementation scripts for the Advanced Falling Object Detection System (AFODS), supporting manuscript vehicles-3989222.

Cyclist safety risk assessment application for Austin using Random Forest classification with SMOTE for class balancing. Interactive Streamlit web app predicts cycling incident severity based on time, location, and environmental conditions to promote safer cycling practices.

Data analysis and interactive visualizations to evaluate young driver safety.

Empirical Bayes before-after study to evaluate ATSC safety effectiveness

Can features of the built environment be used to predict the locations of pedestrian crashes at the street level in Uptown Charlotte, North Carolina?

Orange County Historical Traffic Collisions (OCSWITRS)

Application of ML algorithms in the context of traffic safety analysis

Circulate San Diego: Infrastructure Data Explorer (2023)