Título: Estimation of Safety Distance Between Vehicles on Highways Using YOLOv4 from Aerial Images
Autores: Vítor Freitas, Richardson Menezes, Francisco Vidal and Helton Maia
Resumo:
Traffic accidents are among the most worrying problems in modern life, often caused by human operational errors such as inattention, distraction, and misbehavior. Vehicle speed detection and safety distance measurement can help reduce these accidents. In this study, the computational development conducted was based on Convolutional Neural Networks (CNNs) and the You Only Look Once (YOLO) algorithm to detect vehicles from aerial images and calculate the safe distance and the vehicle’s speed on Brazilian highways. The investigation was conducted to model the YOLO algorithm for detecting vehicles in different network architecture configurations. The best results were obtained with the YOLO Full-608, reaching a mean Average Precision (mAP) of 97.44%. Additional computer vision approaches have been developed to calculate the speed of the moving vehicle and the safe distance between them. Therefore, the developed system allows that, based on detecting the safe distance between moving vehicles on the highways, accidents are predicted and possibly avoided.
Palavras-chave:
Convolutional Neural Networks, You Only Look Once, Speed Detection.
Páginas: 7
Código DOI: 10.21528/CBIC2021-148
Artigo em pdf: CBIC_2021_paper_148.pdf
Arquivo BibTeX: CBIC_2021_148.bib