Universitas Dian Nuswantoro conducted research on a traffic violation detection system for helmet use among motorcyclists. The activity addressed traffic safety and helmet non-compliance, which the source identifies as contributing to accident severity. Researchers participating in the activity were Udayanti, Erika Devi; Firdausillah, Fahri; Affandy, Affandy; Kartikadarma, Etika; and Iksan, Nur.
The research developed a hybrid YOLO-LSTM model. YOLO was used for real-time object detection, while LSTM was used for sequence prediction. The approach combined spatial detection with temporal sequence analysis and was designed to support robust performance across diverse scenarios. The study responded to difficulties experienced by existing detection systems under varying environmental conditions and high traffic volumes. More than 1,100 images captured under varying lighting and weather conditions were used to train and validate the designed model.
The proposed system achieved a mean average precision of 0.777 and demonstrated superior accuracy and recall compared with traditional YOLO-based models. The research also reported improved detection rates in challenging environments. Its findings indicate that hybrid deep learning models can enhance automated traffic enforcement and contribute to safer roads. The activity therefore presents UDINUS research addressing traffic safety and helmet-use enforcement through an applied detection system. The source is too brief to provide a longer account without adding information not explicitly stated.