Computer Vision-Based Hand Baggage Inspection System

Abstract

Object detection is a crucial task in computer vision. It brings two essential tasks, classifying and locating the object in the image or video. This task is performed by specialized human operators, which requires the knowledge and experience for reliable detection. However, the high volume of work often drives the operators to exhaustion, causing a low prevalence of targets. This work aims to build a model using a Deep Neural Network (Darknet) and a single pass object detection method (YOLOv4) for the detection of handguns, shurikens, and blade shavings in hand luggage, using the data set GDXray. The YOLO algorithm model identified different items that could impact security, which could be a tool that makes passenger boarding faster and safer.