A more advanced deep learning architecture that aims to improve in situ driver safety by perceiving the driving behavior and monitoring the surrounding road environment will be proposed in this study. It will employ data provided by in-car cameras and sensors to extract critical spatial features like nearby vehicles, pedestrians, and road signs in addition to temporal patterns reflecting vehicle movement, speed variations, and proximity to potential hazards. The framework integrates spatial and temporal data to identify unsafe driving behaviors such as lane deviations, abrupt speed changes, and unsafe following distances. It further predicts future risks and provides real-time alerts to the driver to prevent accidents. Extensive testing on real-world datasets demonstrates that the system significantly improves hazard detection accuracy, driver behavior monitoring, and performance with low latency, making it a highly practical solution for enhancing safety in modern vehicles. The system plays a vital role in promoting safer roadways and reducing the likelihood of accidents because it can process data in real time.
Hybrid Deep Learning for Proactive Driver Risk Prediction and Safety Enhancement
20.01.2025
411049 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch