Jade Andrie Rosales, Jose Glen Samson, Christian Maderazo
In recent years, the number of motor vehicle accidents has been increasing, causing significant morbidity and mortality worldwide. This study proposes a machine learning model for detecting motor vehicle accidents using the YOLOv8 algorithm to address this issue. The proposed model analyzes video footage captured by cameras installed on roadside infrastructure to detect vehicle accidents. The study uses a publicly available dataset of vehicle accident videos to train and evaluate the model based on the four classifications: 2wheel accident, 4-wheel accident, and large vehicle accident. The model's performance is evaluated using evaluation metrics such as precision, recall, and mean average precision. The developed model, with a precision of 92.4% and a recall of 78.1%, demonstrates notable potential for real-world applications. The model also achieved a Mean Average Precision (mAP) of 87.6% at IoU 0.5, ensuring a high level of accuracy in identifying accidents promptly. © 2023 IEEE.
University of San Carlos, Department of Computer, Information Sciences, and Mathematics, Cebu, Philippines
