Prohibited Object Detection Utilizing Faster R-CNN in X-RAY Images.

Saved in:

Bibliographic Details
Title:	Prohibited Object Detection Utilizing Faster R-CNN in X-RAY Images.
Authors:	KÖKER, Raşit¹ rkoker@subu.edu.tr, SORGUN, Özcan² ozcansorgun17@gmail.com, KUTLU, Mustafa Çağrı³ mkutlu@subu.edu.tr, DEMİR, Mehmet⁴ mehmetdemir@subu.edu.tr
Source:	Technical Gazette / Tehnički Vjesnik. 2026, Vol. 33 Issue 3, p938-948. 11p.
Subjects:	X-ray imaging, Object recognition algorithms, Deep learning, Object recognition (Computer vision), Machine learning, Radioscopic diagnosis
Abstract:	Security checks play a significant role in reducing risks in various areas such as bus terminals, train stations, and airports. At the heart of these checks are X-ray machines, which play a crucial role in detecting weapons, knives, and other prohibited items. However, the increasing volume of goods being transported poses a significant problem: reliance on manual inspection of X-ray security images is fraught with human-induced weaknesses such as negligence, fatigue, and environmental distractions, leading to potential security breaches. This paper proposes an automated solution to address these issues by using the Faster R-CNN deep learning algorithm for detecting prohibited objects in X-ray images. The study utilises the SIXray dataset, comprising X-ray images of prohibited objects across six different categories obtained from various metro stations. The research utilises deep learning-based object detection models to identify five different types of prohibited items using Faster RCNN models (ResNet50, ResNet101, ResNet152, and Inception ResNet v2). These models were trained through transfer learning. The aim is to determine the most efficient model in terms of speed and accuracy for classifying and detecting prohibited objects in X-ray images. The evaluation revealed that the Faster R-CNN ResNet101 model, with a success rate of 90.6% at 0.5 IoU, is superior in terms of accurate classification and object location determination, as evidenced by its precision and recall metrics, mAP values, and image analyses. This comparison with YOLOv technologies revealed that the Faster R-CNN Resnet101 model is superior. This study highlights the potential of automated systems to improve security protocols, reduce costs, and speed up processing times in X-ray security screening. [ABSTRACT FROM AUTHOR]
	Copyright of Technical Gazette / Tehnički Vjesnik is the property of Tehnicki Vjesnik and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Security checks play a significant role in reducing risks in various areas such as bus terminals, train stations, and airports. At the heart of these checks are X-ray machines, which play a crucial role in detecting weapons, knives, and other prohibited items. However, the increasing volume of goods being transported poses a significant problem: reliance on manual inspection of X-ray security images is fraught with human-induced weaknesses such as negligence, fatigue, and environmental distractions, leading to potential security breaches. This paper proposes an automated solution to address these issues by using the Faster R-CNN deep learning algorithm for detecting prohibited objects in X-ray images. The study utilises the SIXray dataset, comprising X-ray images of prohibited objects across six different categories obtained from various metro stations. The research utilises deep learning-based object detection models to identify five different types of prohibited items using Faster RCNN models (ResNet50, ResNet101, ResNet152, and Inception ResNet v2). These models were trained through transfer learning. The aim is to determine the most efficient model in terms of speed and accuracy for classifying and detecting prohibited objects in X-ray images. The evaluation revealed that the Faster R-CNN ResNet101 model, with a success rate of 90.6% at 0.5 IoU, is superior in terms of accurate classification and object location determination, as evidenced by its precision and recall metrics, mAP values, and image analyses. This comparison with YOLOv technologies revealed that the Faster R-CNN Resnet101 model is superior. This study highlights the potential of automated systems to improve security protocols, reduce costs, and speed up processing times in X-ray security screening. [ABSTRACT FROM AUTHOR]
ISSN:	13303651
DOI:	10.17559/TV-20250114002266