APPLICATION OF PROHIBITED OBJECT DETECTION IN AIRPORT BAGGAGE:A hybrid approach with YOLO for gun and knife Identification

Resumo

 https://doi.org/10.5281/zenodo.20187690

The accelerated growth of air transport in recent decades has brought increasing challenges to airport security, especially regarding the detection of unauthorized objects in baggage. Traditional methods—based on X-ray devices and the visual analysis performed by operators—present important limitations, such as human error and fatigue. Seeking to mitigate these flaws, this work proposed and comparatively evaluated different machine learning (ML) models to automatically identify unauthorized objects in X-ray images. The methodology included the collection and pre-processing of the images, the training of the YOLO model, and the evaluation of its performance using metrics such as accuracy, precision, and recall. The results demonstrated that this approach proved to be an agile and reliable solution for identifying threats. This study highlights the potential of artificial intelligence (AI) as a complementary tool in risk prevention in critical environments. The proposal aims to reduce dependence on human examination, increase screening efficiency, and can be integrated into automated inspection systems, contributing to the modernization of processes in the face of the continuous growth in passenger flow.