IoT-based Intelligent Detection System for Pedestrian Safety Enhancement

Pedestrian safety is of utmost importance in developing Autonomous Vehicles (AVs). Accurate detection of pedestrians could be a challenging task to provide safe vehicle and pedestrian interaction. This paper proposes an IoT-based intelligent detection system that employs images captured by a stereo camera mounted on the AV which is called the ETH dataset to enhance pedestrian safety. The proposed system incorporates two deep learning-based models YOLOv5 and YOLOv8 aiming at safe interaction between AVs and pedestrians. The obtained results demonstrate that YOLOv8 achieves higher accuracy and realtime detection capabilities compared to YOLOv5. This advancement enhances safety, particularly in complex and high-density environments, thereby efficiently improving the practicality and reliability of AVs.