SIXray: A Large-Scale Security Inspection X-Ray Benchmark for Prohibited Item Discovery in Overlapping Images

Abstract

In this paper, we present a large-scale dataset and establish a baseline for prohibited item discovery in Security Inspection X-ray images. Our dataset, named SIXray, consists of 1,059,231 X-ray images, in which 6 classes of 8,929 prohibited items are manually annotated. It raises a brand new challenge of overlapping image data, meanwhile shares the same properties with existing datasets, including complex yet meaningless contexts and class imbalance.

An approach for adaptive automatic threat recognition within 3D computed tomography images for baggage security screening

“…In recent years, 2D X-ray image based threat object recognition has been extensively studied [1,2,24], although even the use of multiple view X-ray suffers from the challenges of object recognition under varying orientation and inter-object occlusion. This can be addressed by 3D X-ray CT imaging which provides abundant information as a 3D volume comprised of successive, parallel X-ray image slices [14].…”

On the Use of Deep Learning for the Detection of Firearms in X-ray Baggage Security Imagery

“…Methods such as Faster R-CNN [11], R-FCN [18], and YOLOv2 [19], achieves maximal 0.885 and 0.974 mean Average Precision (mAP) over six-class object detection {Firearm, Firearm Component, Ceramic Knife, Laptop, Camera, Knife} and and two-class firearm detection {Firearm, Firearm Component} problems respectively. Here, we further investigate this performance with our evaluation on X-ray security imagery on larger dataset taken from different source, namely Durham University Full Two-Class and Security Inspection X-ray images [10]. Denoted as Dbf2 [9] and SIXray [10] respectively, it provide more inter-occluding Xray security imagery examples with large variations in pose, scale and item construction.…”

“…Here, we further investigate this performance with our evaluation on X-ray security imagery on larger dataset taken from different source, namely Durham University Full Two-Class and Security Inspection X-ray images [10]. Denoted as Dbf2 [9] and SIXray [10] respectively, it provide more inter-occluding Xray security imagery examples with large variations in pose, scale and item construction.…”

“…Figure 2A depicts exemplary images from this dataset. SIXray is public dataset [10] for prohibited item detection in security inspection X-ray images. These images are collected from several metro sub-way stations, using a Nuctech dualenergy X-ray scanner, where the distribution of the general baggage/parcel items corresponds to stream-of-commerce occurrence.…”

“…localize) prohibited items within X-ray baggage security imagery whilst determining the varying performance of these approaches on different datasets collected for this study. In this approach, we fully utilize large-scale X-ray security imagery provided by [9] and [10]. Denoted as Dbf2 [9] and SIXray [10] respectively, we aim to provide an insight into baseline performance for three CNN architectural variants:-Faster R-CNN [11], Mask R-CNN [12] and RetinaNet [13].…”