考虑样本不平衡的X光安检图像违禁品分类方法

冯霞; 魏新坤; 刘才华; 赫鑫宇

doi:10.13700/j.bh.1001-5965.2022.0095

考虑样本不平衡的X光安检图像违禁品分类方法

doi: 10.13700/j.bh.1001-5965.2022.0095

冯霞^{1, 2},
魏新坤^{1, 2},
刘才华^{1, 2, ,},
赫鑫宇^{1, 2}

1.
中国民航大学计算机科学与技术学院，天津 300300
2.
民航智慧机场理论与系统重点实验室，天津 300300

基金项目: 天津市教委科研计划(2021KJ037)；中央高校基本科研业务费专项资金(3122021052)

详细信息

通讯作者:
E-mail：chliu@cauc.edu.cn

中图分类号: TP391.4
计量
- 文章访问数: 1548
- HTML全文浏览量: 84
- PDF下载量: 34
- 被引次数: 0
出版历程
- 收稿日期: 2022-02-28
- 录用日期: 2022-05-29
- 网络出版日期: 2022-08-22
- 整期出版日期: 2023-12-29

Contraband classification method for X-ray security images considering sample imbalance

FENG Xia^{1, 2},
WEI Xinkun^{1, 2},
LIU Caihua^{1, 2
, ,},
HE Xinyu^{1, 2}

1.
School of Computer Science and Technology，Civil Aviation University of China，Tianjin 300300，China
2.
Key Laboratory of Intelligent Airport Theory and System，CAAC，Tianjin 300300，China

Funds: Scientific Research Project of Tianjin Educational Committee (2021KJ037); The Fundamental Research Funds for the Central Universities (3122021052)

More Information

Corresponding author: E-mail：chliu@cauc.edu.cn

摘要

摘要:
X光安检图像违禁品分类被广泛应用于协助维护航空和运输安全。针对X光安检图像中违禁品尺度不一、存在困难样本及旅客行李安检固有的正负样本不均衡等问题，提出一种端到端的考虑样本不平衡的X光安检图像违禁品分类方法。采用多尺度特征提取网络捕获尺度不一的多类型违禁品特征，通过特征融合模块提升模型对图像边缘和纹理特征的表达能力，基于代价敏感思想设计损失函数，解决数据集不平衡问题，并提高困难样本分类精准度。在公开数据集SIXray上构建的子集实验结果表明：所提方法相较于端到端分类模型，平均AP指标值提升了4.5%，特别是对剪刀等难分类样本，AP指标值都有显著的提升效果。
- 违禁品分类 /
- 样本不平衡 /
- X光图像 /
- 多尺度 /
- 困难样本分类 /
- 代价敏感
Abstract:
X-ray security image contraband classification is widely used to assist in maintaining aviation and transportation security. This paper suggests an end-to-end X-ray security inspection image classification method that takes sample imbalance into account in order to address the issues of different scales of contraband in X-ray images, challenging samples, and unbalanced positive and negative samples inherent in passenger baggage security inspection. The feature fusion module is used to enhance the model’s ability to express picture edge and texture features while the multi-scale feature extraction network is used to capture the features of numerous sorts of illegal goods with various scales. Based on the cost-sensitive idea, the loss function is designed to solve the problem of dataset imbalance, and improve the classification accuracy of difficult samples.The experimental results of the subset constructed on the public dataset SIXray show that the proposed method improves the mean AP index by 4.5% compared with the current optimal end-to-end classification model, especially for hard-to-classify samples such as scissors, the AP index has a significant improvement effect.
- contraband classification /
- sample imbalance /
- X-ray images /
- multi-scale /
- difficult sample classification /
- cost-sensitive

HTML全文

图 1 模型结构

Figure 1. Model structure

下载: 全尺寸图片幻灯片

图 2 多尺度特征提取网络

Figure 2. Multiscale feature extraction network

下载: 全尺寸图片幻灯片

图 3 违禁品AP指标与代价敏感参数的关系

Figure 3. Relationship between AP index and cost sensitive parameters of contraband

下载: 全尺寸图片幻灯片

图 4 违禁品AP指标与正负样本权重因子的关系

Figure 4. Relationship between AP index and positive and negative sample weight factor of contraband

下载: 全尺寸图片幻灯片

图 5 违禁品AP指标与难分类样本权重因子的关系

Figure 5. Relationship between AP index and difficult-to-score sample weight factor of contraband

下载: 全尺寸图片幻灯片

表 1 不同模型分类性能对比实验结果

Table 1. Comparative experiment results of classification performance of different models %

方法	AP指标值					平均AP指标值
方法	刀	枪	扳手	钳子	剪刀	平均AP指标值
ResNet101	84.2	87.7	69.3	85.3	60.4	77.4
Inception-V3	83.8	90.1	68.1	84.5	58.7	77.0
RFBNet	72.9	90.5	64.9	77.3	68.6	74.8
ACMNet	80.2	91.5	83.6	85.9	80.3	84.3
Cascade R-CNN	80.4	87.4	82.4	86.4	86.3	84.6
ResNet101+CHR	87.2	85.5	71.2	88.3	64.7	79.4
本文	90.5	86.5	80.5	97.3	89.4	88.8

下载: 导出CSV

表 2 消融实验结果

Table 2. Ablation experiment results

骨干网络	特征融合模块	多尺度残差学习模块	FCB Loss	平均AP 指标值/%
√				77.4
√	√			79.4
√	√	√		83.3
√	√	√	√	88.8

下载: 导出CSV

参考文献(19)

[1]	HEITZ G, CHECHIK G. Object separation in X-ray image sets[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2010: 2093-2100.
[2]	MERY D. Automated detection in complex objects using a tracking algorithm in multiple X-ray views[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2011: 12173892.
[3]	HASSAN T, WERGHI N. Trainable structure tensors for autonomous baggage threat detection under extreme occlusion[C]//Proceedings of the Asian Conference on Computer Vision. Berlin: Springer, 2020: 257-273.
[4]	SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2015: 1-9.
[5]	AKÇAY S, KUNDEGORSK I M E, DEVEREUX M, et al. Transfer learning using convolutional neural networks for object classification within X-ray baggage security imagery[C]//Proceedings of the IEEE International Conference on Image Processing. Piscataway: IEEE Press, 2016: 1057-1061.
[6]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2016: 779-788.
[7]	张友康, 苏志刚, 张海刚, 等. X光安检图像多尺度违禁品检测[J]. 信号处理, 2020, 36(7): 1096-1106. doi: 10.16798/j.issn.1003-0530.2020.07.008 ZHANG Y K, SU Z G, ZHANG H G, et al. X-ray security images multiscale contraband detection[J]. Signal Processing, 2020, 36(7): 1096-1106(in Chinese). doi: 10.16798/j.issn.1003-0530.2020.07.008
[8]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: Single shot multibox detector[C]//Proceedings of the European Conference on Computer Vision. Berlin: Springer, 2016: 21-37.
[9]	MIAO C, XIE L, WAN F, et al. SIXray: A large-scale security in spection X-ray benchmark for prohibited item discovery in overlapping images[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2019: 2114-2123.
[10]	HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2016: 770-778.
[11]	WEBB T W, BHOWMIK N, GAUS Y F A, et al. Operationalizing convolutional neural network architectures for prohibited object detection in X-ray imagery[C]//Proceedings of the IEEE International Conference on Machine Learning and Applications. Piscataway: IEEE Press, 2021: 610-615.
[12]	CAI Z, VASCONCELOS N. Cascade R-CNN: High quality object detection and instance segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(5): 1483-1498.
[13]	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]//Proceedings of the IEEE International Conference on Computer Vision. Piscataway: IEEE Press, 2017: 2999-3007.
[14]	闫明松, 周志华. 代价敏感分类算法的实验比较[J]. 模式识别与人工智能, 2005, 18(5): 8. YAN M S, ZHOU Z H. An empirical comparative study of cost-sensitive classification algorithms[J]. Pattern Recognition and Artificial Intelligence, 2005, 18(5): 8(in Chinese).
[15]	GAO S, CHENG M M, ZHAO K, et al. Res2Net: A new multi-scale backbone architecture[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(2): 652-662. doi: 10.1109/TPAMI.2019.2938758
[16]	CUI Y, JIA M, LIN T Y, et al. Class-balanced loss based on effective number of samples[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2019: 19262778.
[17]	EVERINGHAM M, ESLAMI S, VAN GOOL L, et al. The PASCAL visual object classes challenge: A retrospective[J]. International Journal of Computer Vision, 2015, 111(1): 98-136. doi: 10.1007/s11263-014-0733-5
[18]	SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2016: 2818-2826.
[19]	LIU S, HUANG D, WANG Y H. Receptive field block net for accurate and fast object detection[C]∥Proceedings of the European Conference on Computer Vision. Berlin: Springer, 2018: 404-419.