Project R-14406

Enhancing Road Safety through UAV captured Aerial Footage and Deep Learning: Object Detection, Risk Prediction, and Real-time Analysis. (Research)

The proposed research aims to improve road safety and enhance the overall driving experience by integrating computer vision, deep learning techniques, and Unmanned Aerial Vehicles (UAVs). The study focuses on several key areas: exploring instance segmentation for categorized vehicle detection, developing real-time analysis tools for traffic safety using multi-source data, enhancing pedestrian detection through the integration of spatiotemporal information, optimizing surrogate safety indicators for predicting road accident risk and severity, developing an automated flight planning approach for UAV access restriction, and investigating the impact of the built environment on road safety. The anticipated outcomes include a comparative analysis of object detection algorithms, an automated real-time road safety improvement system, a robust pedestrian detection system, a modified accident damage prediction model, an automated flight planning system, and an improved understanding of the urban environment's role in road accidents. The findings will enhance road safety, inform accident prevention strategies, support urban planning, and improve transportation management.

19 December 2023 - 31 October 2027