Résultats du processus de synthèse de l'article arXiv : 2303.11223v1

Résumé : Cycling is an increasingly popular method of transportation for sustainability and health benefits. However, cyclists face growing risks, especially when encountering semi-trailer trucks. This study aims to reduce the number of truck-cyclist collisions, which are often caused by semi-trailer trucks making right-hook turns and poor driver attention to blind spots. To achieve this, we designed a visual-based blind spot warning system that can detect cyclists for semi-trailer truck drivers using deep learning. First, several greater than 90% mAP cyclist detection models, such as the EfficientDet Lite 1 and SSD MobileNetV2, were created using state-of-the-art lightweight deep learning architectures fine-tuned on a newly proposed cyclist image dataset composed of a diverse set of over 20,000 images. Next, the object detection model was deployed onto a Google Coral Dev Board mini-computer with a camera module and analyzed for speed, reaching inference times as low as 15 milliseconds. Lastly, the end-to-end blind spot cyclist detection device was tested in real-time to model traffic scenarios and analyzed further for performance and feasibility. We concluded that this portable blind spot alert device can accurately and quickly detect cyclists and have the potential to significantly improve cyclist safety. Future studies could determine the feasibility of the proposed device in the trucking industry and improvements to cyclist safety over time.