Enhancing Real-Time Detection Transformer (RT-DETR) for Handgun Detection on Nvidia Jetson

Bustamante L.A.; Gutiérrez J.C. · CLEI Eletronic Journal (CLEIej) (2025) · DOI: 10.19153/cleiej.28.3.5

Resumen

Recent studies have highlighted the rise in violence and criminal activities, particularly those involving firearms. In response to the growing demand for effective firearm detection systems in public safety applications, this paper explores advancements in real-time object detection using Transformer-based models. Building on the RT-DETR architecture and its latest version, RT-DETR v2, we introduce modifications such as the Bidirectional Feature Pyramid Network (BiFPN), achieving an improvement of over 17% in small object detection specifically firearms compared to previous work. Additionally, we employ dynamic batch processing to maximize computational efficiency and resource utilization on edge devices like the Nvidia Jetson AGX Xavier, enabling efficient real-time deployment. We also compare our model with state-of-the-art alternatives such as YOLOv10, demonstrating the superior accuracy and performance of Transformer-based models. For the comparative study, we use a benchmark that includes three datasets with challenging conditions. The code and trained models are available at https://github.com/labt1/GunDetection-RTDETR.

Instituciones: Universidad Nacional de San Agustín de Arequipa