Occluded Prohibited Items Detection: An X-ray Security Inspection Benchmark and De-occlusion Attention Module

Wei, Yanlu; Tao, Renshuai; Wu, Zhangjie; Ma, Yuqing; Zhang, Libo; Liu, Xianglong

doi:10.1145/3394171.3413828

Cited by 148 publications

(170 citation statements)

References 27 publications

Supporting

Mentioning

128

Contrasting

Order By: Relevance

“…However, deep learning frameworks are still vulnerable to extreme occlusion, clutter, and diverse scanner specifications. Although, recent developments in recognizing baggage threats managed to address occlusion to some extent [ 5 , 13 , 14 ]. However, these frameworks are either tested on a single dataset [ 13 , 14 ] or they require extensive (parameter) tuning for different scanner specifications [ 5 ].…”

Section: Related Workmentioning

confidence: 99%

“…Although, recent developments in recognizing baggage threats managed to address occlusion to some extent [ 5 , 13 , 14 ]. However, these frameworks are either tested on a single dataset [ 13 , 14 ] or they require extensive (parameter) tuning for different scanner specifications [ 5 ]. Furthermore, to the best of our knowledge, there is no mechanism (based on meta-learning [ 16 ] or meta-transfer learning [ 17 ]) to extend the capacity of these frameworks to generalize well across diverse ranging scanners without an explicit retraining process.…”

Section: Related Workmentioning

confidence: 99%

“…Furthermore, they compared Faster R-CNN [ 42 ], sliding-window based CNN (SW-CNN), region-based fully convolutional networks (R-FCN) [ 50 ], and YOLOv2 [ 40 ] for recognizing occluded contraband items from the X-ray imagery. More recently, Wei et al [ 13 ] proposed De-occlusion Attention Module (DOAM) module that can be integrated with the deep object detectors to recognize occluded threatening items. DOAM was thoroughly validated on a large-scale Occluded Prohibited Items X-ray (OPIXray) dataset, released publicly in [ 13 ].…”

Section: Related Workmentioning

confidence: 99%

“…However, due to the inherent differences between the X-ray and the RGB scans, these object detectors do not work well for identifying the baggage threats (via X-ray imagery) [ 8 , 9 , 10 ], especially in extreme concealment and cluttered scenarios [ 5 , 11 ]. To overcome this, researchers have developed exclusive frameworks for detecting and classifying baggage threats from the X-ray scans [ 12 , 13 , 14 ]. These frameworks can well recognize the visible and partially occluded baggage threats from the X-ray scans [ 5 , 11 , 13 ].…”

Section: Introductionmentioning

confidence: 99%

“…To overcome this, researchers have developed exclusive frameworks for detecting and classifying baggage threats from the X-ray scans [ 12 , 13 , 14 ]. These frameworks can well recognize the visible and partially occluded baggage threats from the X-ray scans [ 5 , 11 , 13 ]. However, they are still vulnerable towards recognizing the extremely cluttered, concealed, and occluded objects [ 5 , 9 ] like, for example, the guns in Figure 1 A–F,I, the knives in Figure 1 F–I, and the wrenches in Figure 1 F.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Meta-Transfer Learning Driven Tensor-Shot Detector for the Autonomous Localization and Recognition of Concealed Baggage Threats

Hassan

Shafay

Akçay

et al. 2020

Sensors

View full text Add to dashboard Cite

Screening baggage against potential threats has become one of the prime aviation security concerns all over the world, where manual detection of prohibited items is a time-consuming and hectic process. Many researchers have developed autonomous systems to recognize baggage threats using security X-ray scans. However, all of these frameworks are vulnerable against screening cluttered and concealed contraband items. Furthermore, to the best of our knowledge, no framework possesses the capacity to recognize baggage threats across multiple scanner specifications without an explicit retraining process. To overcome this, we present a novel meta-transfer learning-driven tensor-shot detector that decomposes the candidate scan into dual-energy tensors and employs a meta-one-shot classification backbone to recognize and localize the cluttered baggage threats. In addition, the proposed detection framework can be well-generalized to multiple scanner specifications due to its capacity to generate object proposals from the unified tensor maps rather than diversified raw scans. We have rigorously evaluated the proposed tensor-shot detector on the publicly available SIXray and GDXray datasets (containing a cumulative of 1,067,381 grayscale and colored baggage X-ray scans). On the SIXray dataset, the proposed framework achieved a mean average precision (mAP) of 0.6457, and on the GDXray dataset, it achieved the precision and F1 score of 0.9441 and 0.9598, respectively. Furthermore, it outperforms state-of-the-art frameworks by 8.03% in terms of mAP, 1.49% in terms of precision, and 0.573% in terms of F1 on the SIXray and GDXray dataset, respectively.

show abstract