基于轻量化卷积和SCAM改进的X光违禁品检测

doi:10.3778/j.issn.1673-9418.2404059

摘要/Abstract

摘要： 针对X光违禁品图像目标重叠遮挡、关键特征信息提取困难和复杂背景干扰等问题，提出了多分支轻量化卷积和注意力机制改进的X光违禁品检测模型。所提模型在主干网络设计空间和通道重构注意力机制（SCAM），通过对特征图在通道和空间上重组，区分特征图冗余信息和非冗余信息，加强关键特征提取并抑制背景干扰，提升模型面对复杂场景检测违禁品的能力；提出多分支轻量化卷积结构（MLCB），采用轻量化双分支和信息补偿分支共同处理特征图，降低模型参数量并防止模型预测精度下滑，提升运行效率；同时融合最小交并比（MPDIoU）损失函数和软非极大值抑制（Soft NMS）替换完全交并比（CIoU）边框回归损失函数，通过定义更加全面的交并比方式，缓解边框回归重合情况下难以优化问题，改善违禁品重叠遮挡造成的易漏检误检问题。所提模型在OPIXray、HIXray与 SIXray三个数据集上进行验证，mAP0.50分别达到了95.7%、83.7%和95.3%。实验结果表明所提方法在计算量较小的情况下，具备高精准度和强鲁棒性，可以有效解决重叠遮挡和漏报、误报等问题。

关键词: X光图像, 违禁品检测, 空间和通道重构, 多分支轻量化卷积, 损失函数

Abstract: To resolve the problems of high overlap, occlusion, difficulty in extracting key feature information, and complex background interference in X-ray contraband images, an improved X-ray contraband detection model based on multi-branch lightweight convolution and attention mechanism is proposed. The proposed model designs the backbone network with the spatial-channel attention mechanism (SCAM), which distinguishes redundant and non-redundant information in feature maps by reorganizing them in the channel and spatial dimensions, enhancing the extraction of key features and suppressing background interference, and also improving the ability of the model to detect contraband in complex scenes. A multi-branch lightweight convolution structure (MLCB) is proposed, which uses a lightweight dual-branch and information compensation branch to process feature maps to reduce the number of model parameters, so as to improve the operation efficiency. In addition, the minimum point distance intersection over union (MPDIoU) loss function and soft non-maximum suppression (Soft NMS) are integrated to replace the complete intersection over union (CIoU) loss function. By defining a more comprehensive intersection over union method, it alleviates the difficulty in optimization when bounding boxes overlap, and improves the problems of false positive and false negative caused by overlap of contraband. The proposed model is verified on OPIXray, HIXray and SIXray datasets, and mAP0.50 reaches 95.7%, 83.7% and 95.3%, respectively. The results show that the proposed method has high accuracy and strong robustness with a small amount of calculation, and it can solve the problems of overlap and occlusion as well as missed and false detections effectively.

Key words: X-ray images, contraband detection, spatial and channel reconstruction, multi-branch light convolution, loss function

左景, 石洋宇, 卢树华. 基于轻量化卷积和SCAM改进的X光违禁品检测[J]. 计算机科学与探索, 2025, 19(6): 1598-1610.

ZUO Jing, SHI Yangyu, LU Shuhua. Improved X-ray Prohibited Items Detection Method Based on Lightweight Convolution Blocks and SCAM[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(6): 1598-1610.

参考文献

[1] YANG C, HUANG Z H, WANG N Y. QueryDet: cascaded sparse query for accelerating high-resolution small object detection[C]//Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2022: 13658-13667.
[2] LIANG K J, SIGMAN J B, SPELL G P, et al. Toward automatic threat recognition for airport X-ray baggage screening with deep convolutional object detection[EB/OL]. [2024-01-19]. https://arxiv.org/abs/1912.06329.
[3] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision. Cham: Springer, 2016: 21-37.
[4] BASTAN M, BYEON W, BREUEL T. Object recognition in multi-view dual energy X-ray images[C]//Proceedings of the 2013 British Machine Vision Conference, 2013.
[5] ZHENG Q H, SAPONARA S, TIAN X Y, et al. A real-time constellation image classification method of wireless communication signals based on the lightweight network MobileViT[J]. Cognitive Neurodynamics, 2024, 18(2): 659-671.
[6] ZHENG Q H, TIAN X Y, YU Z G, et al. MobileRaT: a lightweight radio transformer method for automatic modulation classification in drone communication systems[J]. Drones, 2023, 7(10): 596.
[7] AKCAY S, BRECKON T P. An evaluation of region based object detection strategies within X-ray baggage security imagery[C]//Proceedings of the 2017 IEEE International Conference on Image Processing. Piscataway: IEEE, 2017: 1337- 1341.
[8] GRIFFIN L D, CALDWELL M, ANDREWS J T A, et al. “Unexpected item in the bagging area”: anomaly detection in X-ray security images[J]. IEEE Transactions on Information Forensics and Security, 2019, 14(6): 1539-1553.
[9] MIAO C J, XIE L X, WAN F, et al. SIXray: a large-scale security inspection X-ray benchmark for prohibited item discovery in overlapping images[C]//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2019: 2114-2123.
[10] BHOWMIK N, GAUS Y F A, AKCAY S, et al. On the impact of object and sub-component level segmentation strategies for supervised anomaly detection within X-ray security imagery[C]//Proceedings of the 2019 18th IEEE International Conference on Machine Learning and Applications. Piscataway: IEEE, 2019: 986-991.
[11] HASSAN T, AKCAY S, BENNAMOUN M, et al. Cascaded structure tensor framework for robust identification of heavily occluded baggage items from X-ray scans[EB/OL]. [2024-01-19]. https://arxiv.org/abs/2004.06780.
[12] TAO R S, WEI Y L, JIANG X J, et al. Towards real-world X-ray security inspection: a high-quality benchmark and lateral inhibition module for prohibited items detection[C]//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2021: 10903-10912.
[13] LIU D M, LIU J C, YUAN P X, et al. A lightweight dangerous liquid detection method based on depthwise separable convolution for X-ray security inspection[J]. Computational Intelligence and Neuroscience, 2022(1): 5371350.
[14] XU B, GAO B, LI Y H. Improved small object detection algorithm based on YOLOv5[J]. IEEE Intelligent Systems, 2024, 39(5): 57-65.
[15] MA S L, XU Y, MA S L, et al. MPDIoU: a loss for efficient and accurate bounding box regression[EB/OL]. [2024-01-19]. https://arxiv.org/abs/2307.07662.
[16] CHEN Y S, WANG L Z, DING B J, et al. Automated Alzheimer??s disease classification using deep learning models with Soft-NMS and improved ResNet50 integration[J]. Journal of Radiation Research and Applied Sciences, 2024, 17(1): 100782.
[17] BODLA N, SINGH B, CHELLAPPA R, et al. Soft-NMS: improving object detection with one line of code[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2017: 5562-5570.
[18] JOCHER G. Ultralytics YOLOv8[EB/OL]. [2024-01-19]. https:// github.com/ultralytics/ultralytics.
[19] CHEN J R, KAO S H, HE H, et al. Run, don’t walk: chasing higher FLOPS for faster neural networks[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2023: 12021-12031.
[20] HOU Q B, ZHOU D Q, FENG J S. Coordinate attention for efficient mobile network design[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 13708-13717.
[21] LI P W, ZHANG L J, ZHOU X D, et al. Attention based network with DA-loss for X-ray contraband automatic detection[C]//Proceedings of the 2023 IEEE International Conference on Multimedia and Expo. Piscataway: IEEE, 2023: 2237-2242.
[22] WEI Y L, TAO R S, WU Z J, et al. Occluded prohibited items detection: an X-ray security inspection benchmark and de-occlusion attention module[C]//Proceedings of the 28th ACM International Conference on Multimedia. New York: ACM, 2020: 138-146.
[23] LI J F, WEN Y, HE L H. SCConv: spatial and channel reconstruction convolution for feature redundancy[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2023: 6153- 6162.
[24] YU J H, JIANG Y N, WANG Z Y, et al. UnitBox: an advanced object detection network[C]//Proceedings of the 24th ACM International Conference on Multimedia. New York: ACM, 2016: 516-520.
[25] ZHENG Z H, WANG P, LIU W, et al. Distance-IoU loss: faster and better learning for bounding box regression[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(7): 12993-13000.
[26] REZATOFIGHI H, TSOI N, GWAK J, et al. Generalized intersection over union: a metric and a loss for bounding box regression[C]//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2019: 658-666.
[27] ZHANG Y F, REN W Q, ZHANG Z, et al. Focal and efficient IOU loss for accurate bounding box regression[J]. Neurocomputing, 2022, 506: 146-157.
[28] ZHANG Y, ZHUO L, MA C J, et al. CTA-FPN: channel-target attention feature pyramid network for prohibited object detection in X-ray images[J]. Sensing and Imaging, 2023, 24(1): 14.
[29] 孙嘉傲, 董乙杉, 郭靖圆, 等. 自适应与多尺度特征融合的X光违禁品检测[J]. 计算机工程与应用, 2024, 60(2): 96-102.
SUN J A, DONG Y S, GUO J Y, et al. Detection of X-ray contraband by adaptive and multi-scale feature fusion[J]. Computer Engineering and Applications, 2024, 60(2): 96-102.
[30] LI C Y, LI L, JIANG H L, et al. YOLOv6: a single-stage object detection framework for industrial applications[EB/OL]. [2024-02-20]. https://arxiv.org/abs/2209.02976.
[31] XIANG N, GONG Z H, XU Y, et al. Material-aware path aggregation network and shape decoupled SIoU for X-ray contraband detection[J]. Electronics, 2023, 12(5): 1179.
[32] DAI Y F, CHEN P C. YOLO lightweight contraband detection network using attention mechanism[C]//Proceedings of the 2023 International Conference on Mechatronics Engineering and Artificial Intelligence, 2023: 302-306.
[33] WANG Z S, WANG X H, SHI Y T, et al. Lightweight detection method for X-ray security inspection with occlusion[J]. Sensors, 2024, 24(3): 1002.
[34] HASSAN T, AK?AY S, BENNAMOUN M, et al. Tensor pooling-driven instance segmentation framework for baggage threat recognition[J]. Neural Computing and Applications, 2022, 34(2): 1239-1250.
[35] HAN L, MA C H, LIU Y, et al. SC-YOLOv8: a security check model for the inspection of prohibited items in X-ray images[J]. Electronics, 2023, 12(20): 4208.
[36] LIU W, SUN D G, WANG Y, et al. ABTD-net: autonomous baggage threat detection networks for X-ray images[C]//Proceedings of the 2023 IEEE International Conference on Multimedia and Expo. Piscataway: IEEE, 2023: 1229-1234.
[37] GLENN J, ALEX S, JIRKA B, et al. ultralytics/YOLOv5: v6.1-YOLOv5.P6[EB/OL]. [2024-02-20]. https://github.com/ ultralytics/yolov5.
[38] MA C J, ZHUO L, LI J F, et al. Occluded prohibited object detection in X-ray images with global context-aware multi-scale feature aggregation[J]. Neurocomputing, 2023, 519: 1-16.
[39] TIAN Z, SHEN C H, CHEN H, et al. FCOS: fully convolutional one-stage object detection[C]//Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2019: 9627-9636.
[40] WANG Z S, ZHANG H Y, LIN Z B, et al. Prohibited items detection in baggage security based on improved YOLOv5[C]//Proceedings of the 2022 IEEE 2nd International Conference on Software Engineering and Artificial Intelligence. Piscataway: IEEE, 2022: 20-25.
[41] WANG H Y, WANG W, LIU Y. X-YOLO: a deep learning based toolset with multiple optimization strategies for contraband detection[C]//Proceedings of the 2020 ACM Turing Celebration Conference. New York: ACM, 2020: 127-132.
[42] 成浪, 敬超. 基于改进YOLOv7的X线图像旋转目标检测[J]. 图学学报, 2023, 44(2): 324-334.
CHENG L, JING C. X-ray image rotating object detection based on improved YOLOv7[J]. Journal of Graphics, 2023, 44(2): 324-334.
[43] YU X Y, YUAN W J, WANG A L. X-ray security inspection image dangerous goods detection algorithm based on improved YOLOv4[J]. Electronics, 2023, 12(12): 2644.
[44] ZHANG J K, LIU Y, LV X Q, et al. X-ray image prohibited item detection algorithm based on improved PP-YOLO[J]. Journal of Computers, 2023, 34(4): 53-68.
[45] MADEMLIS I, BATSIS G, CHRYSOCHOOU A A R, et al. Visual inspection for illicit items in X-ray images using deep learning[C]//Proceedings of the 2023 IEEE International Conference on Big Data. Piscataway: IEEE, 2023: 4081- 4089.
[46] NASIM A, VELAYUDHAN D, AHMED A H, et al. Incremental instance segmentation for cluttered baggage threat detection[C]//Proceedings of the 2023 IEEE International Conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications. Piscataway: IEEE, 2023: 1-6.
[47] LU L Y. Improved YOLOv8 detection algorithm in security inspection image[EB/OL]. [2024-02-20]. https://arxiv.org/abs/ 2308.06452.
[48] 穆思奇, 林进健, 汪海泉, 等. 基于改进YOLOv4的X射线图像违禁品检测算法[J]. 兵工学报, 2021, 42(12): 2675-2683.
MU S Q, LIN J J, WANG H Q, et al. An algorithm for detection of prohibited items in X-ray images based on improved YOLOv4[J]. Acta Armamentarii, 2021, 42(12): 2675- 2683.
[49] GEVORGYAN Z. SIoU loss: more powerful learning for bounding box regression[EB/OL]. [2024-02-20]. https://arxiv.org/abs/2205.12740.
[50] TONG Z J, CHEN Y H, XU Z W, et al. Wise-IoU: bounding box regression loss with dynamic focusing mechanism[EB/OL]. [2024-02-20]. https://arxiv.org/abs/2301.10051.
[51] LIU Z H, WANG H, ZHANG S T, et al. NAS-SCAM: neural architecture search-based spatial and channel joint attention module for nuclei semantic segmentation and classification[C]//Proceedings of the 23rd International Conference on Medical Image Computing and Computer Assisted Intervention. Cham: Springer, 2020: 263-272.