Enhanced SSD framework for detecting defects in cigarette appearance using variational Bayesian inference under limited sample conditions

Shichao Wu; Xianzhou Lv; Yingbo Liu; Ming Jiang; Xingxu Li; Dan Jiang; Jing Yu; Yunyu Gong; Rong Jiang; Shichao Wu; Xianzhou Lv; Yingbo Liu; Ming Jiang; Xingxu Li; Dan Jiang; Jing Yu; Yunyu Gong; Rong Jiang

doi:10.3934/mbe.2024145

Mathematical Biosciences and Engineering

2024, Volume 21, Issue 2: 3281-3303. doi: 10.3934/mbe.2024145

Previous Article Next Article

Research article Special Issues

Enhanced SSD framework for detecting defects in cigarette appearance using variational Bayesian inference under limited sample conditions

1.
School of Statistics and Mathematics, Yunnan University of Finance and Economics, Kunming 650221, China
2.
Hongyun Honghe Tobacco (Group) Co., Ltd. Huize Cigarette Factory, Qujing 654200, China
3.
Yunnan Key Laboratory of Service Computing, Kunming 650221, China
^† These authors contributed equally to this work.

Received: 03 December 2023 Revised: 25 January 2024 Accepted: 29 January 2024 Published: 04 February 2024

In high-speed cigarette manufacturing industries, occasional minor cosmetic cigarette defects and a scarcity of samples significantly hinder the rapid and accurate detection of defects. To tackle this challenge, we propose an enhanced single-shot multibox detector (SSD) model that uses variational Bayesian inference for improved detection of tiny defects given sporadic occurrences and limited samples. The enhanced SSD model incorporates a bounded intersection over union (BIoU) loss function to reduce sensitivity to minor deviations and uses exponential linear unit (ELU) and leaky rectified linear unit (ReLU) activation functions to mitigate vanishing gradients and neuron death in deep neural networks. Empirical results show that the enhanced SSD300 and SSD512 models increase the model's detection accuracy mean average precision (mAP) by up to 1.2% for small defects. Ablation studies further reveal that the model's mAP increases by 1.5%, which reduces the computational requirements by 5.92 GFLOPs. The model also shows improved inference in scenarios with limited samples, thus highlighting its effectiveness and applicability in high-speed, precision-oriented cigarette manufacturing industries.
- SSD,
- cigarette appearance defects,
- variational Bayesian inference,
- tiny target detection
Citation: Shichao Wu, Xianzhou Lv, Yingbo Liu, Ming Jiang, Xingxu Li, Dan Jiang, Jing Yu, Yunyu Gong, Rong Jiang. Enhanced SSD framework for detecting defects in cigarette appearance using variational Bayesian inference under limited sample conditions[J]. Mathematical Biosciences and Engineering, 2024, 21(2): 3281-3303. doi: 10.3934/mbe.2024145

Related Papers:

Abstract

In high-speed cigarette manufacturing industries, occasional minor cosmetic cigarette defects and a scarcity of samples significantly hinder the rapid and accurate detection of defects. To tackle this challenge, we propose an enhanced single-shot multibox detector (SSD) model that uses variational Bayesian inference for improved detection of tiny defects given sporadic occurrences and limited samples. The enhanced SSD model incorporates a bounded intersection over union (BIoU) loss function to reduce sensitivity to minor deviations and uses exponential linear unit (ELU) and leaky rectified linear unit (ReLU) activation functions to mitigate vanishing gradients and neuron death in deep neural networks. Empirical results show that the enhanced SSD300 and SSD512 models increase the model's detection accuracy mean average precision (mAP) by up to 1.2% for small defects. Ablation studies further reveal that the model's mAP increases by 1.5%, which reduces the computational requirements by 5.92 GFLOPs. The model also shows improved inference in scenarios with limited samples, thus highlighting its effectiveness and applicability in high-speed, precision-oriented cigarette manufacturing industries.

References

[1]	China Tobacco Machinery, Changde Tobacco Machinery Limited Liability Company, 2022. Available from: https://www.ccdtm.com/info/info/_intr.jsp.
[2]	G.D 121P-20K Cigarette Manufacturing Machines, Gulf Tobacco, 2023. Available from: http://www.cigarettemanufacturingmachines.com/gd-121p-20k-cigarette-manufacturing-machine/.
[3]	121P Double rod cigarette maker, Gulf Tobacco, 2023. Available from: https://www.gidi.it/en/solutions/product/121p.
[4]	Z. Y. Xiao, Research and implementation of cigarette defect detection algorithm, Yunnan Univ., 2018.
[5]	Y. X. Yang, Design and implementation of an image processing based cigarette defect detection method, Yunnan Univ., 2018.
[6]	J. Redmon, S. Divvala, R. Girshick, A. Farhadi, You only look once: Unified, real-time object detection, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2016), 779–788. https://doi.org/10.1109/CVPR.2016.91
[7]	W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C. Fu, et al., Ssd: Single shot multibox detector, in Proceedings of the European Conference on Computer Vision(ECCV), 9905 (2016), 21–37. https://doi.org/10.1007/978-3-319-46448-0_2
[8]	A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, et al., Attention is all you need, Adv. Neural Inform. Process. Syst., 30 (2017). https://doi.org/10.48550/arXiv.1706.03762
[9]	K. Shridhar, F. Laumann, M. Liwicki, Uncertainty estimations by softplus normalization in bayesian convolutional neural networks with variational inference, preprint, arXiv: 1806.05978.
[10]	R. Qu, G. Yuan, J. Liu, H. Zhou, Detection of cigarette appearance defects based on improved SSD model, in Proceedings of the 2021 5th International Conference on Electronic Information Technology and Computer Engineering, (2021), 1148–1153. https://doi.org/10.1145/3501409.3501612
[11]	Z. W. Du, H. Zhou, C. Y. Li, Small object detection based on deep convolutional neural networks: A review, Comput. Sci., 49 (2022), 205–208. https://doi.org/10.11896/jsjkx.220500260 doi: 10.11896/jsjkx.220500260
[12]	H. J. Yang, L. Meng, An improved algorithm for small target detection in aerial photography images based on YOLOv5, Comput. Eng. Sci., 45 (2023), 1063–1070.
[13]	L. Li, M. Li, H. Hu, An algorithm for cigarette capsules defect detection based on lightweight faster rcnn, in 2021 40th Chinese Control Conference (CCC), (2021), 8028–8034. https://doi.org/10.23919/CCC52363.2021.9550392
[14]	E. Kim, J. Lee, H. Jo, K. Na, E. Moon, G. Gweon, et al., SHOMY: Detection of small hazardous objects using the you only look once algorithm, KSII Trans. Int. Inform. Syst. (TIIS), 16 (2022), 2688–2703. https://doi.org/10.3837/tiis.2022.08.012 doi: 10.3837/tiis.2022.08.012
[15]	J. Diers, C. Pigorsch, A survey of methods for automated quality control based on images, Int. J. Comput. Vis., 131 (2023), 2348–2356. https://doi.org/10.1007/s11263-023-01822-w doi: 10.1007/s11263-023-01822-w
[16]	H. Q. Wang, Deep learning-based target detection of cigarette defects, Shenyang Univ. Chem. Techn., 2022. https://doi.org/10.27905/d.cnki.gsghy.2022.000092
[17]	G. W. Yuan, J. C. Liu, R. Qu, H. Zhou, Classification of cigarette appearance defects based on ResNeSt, J. Yunnan Univ. Natural Sci. Edition, 44 (2022), 464–470. https://doi.org/10.7540/j.ynu.20210257 doi: 10.7540/j.ynu.20210257
[18]	H. Y. Liu, G. W. Yuan, L. Yang, K. Liu, H. Zhou, An appearance defect detection method for cigarettes based on C-CenterNet, Electronics, 11 (2022), 2182. https://doi.org/10.3390/electronics11142182 doi: 10.3390/electronics11142182
[19]	Y. L. Li, S. Yang, L. F. Fan, Y. H. Xiong, Q. Zhu, L. H. Zhang, Online inspection of cigarette seam defects based on machine vision, Tobacco Sci. Technol., 56 (2023), 93–98. https://doi.org/10.16135/j.issn1002-0861.2022.0474 doi: 10.16135/j.issn1002-0861.2022.0474
[20]	G. W. Yuan, J. C. Liu, H. Y. Liu, Y. Ma, H. Wu, H. Zhou, Detection of cigarette appearance defects based on improved YOLOv4, Electr. Res. Arch., 31 (2023), 1344–1364. https://doi.org/10.3934/era.2023069. doi: 10.3934/era.2023069
[21]	Y. H. Ma, G. W. Yuan, K. Yue, H. Zhou, CJS-YOLOv5n: A high-performance detection model for cigarette appearance defects, Math. Biosci. Eng., 20 (2023), 17886–17904. https://doi.org/10.3934/mbe.2023795 doi: 10.3934/mbe.2023795
[22]	H. Y. Liu, G. W. Yuan, Detection of cigarette appearance defects based on improved YOLOv5s, Comput. Technol. Dev., 32 (2022), 161–167.
[23]	D. Feng, Z. G. Li, A. M. He, X. Yang, S. Wang, H. Dong, et al., Appearance quality inspection of cigarette products based on local characteristic similarity metric, Tobacco Sci. Technol., 56 (2023), 82–90. https://doi.org/10.16135/j.issn1002-0861.2022.0807 doi: 10.16135/j.issn1002-0861.2022.0807
[24]	R. Qu, Research on cigarette appearance defect detection based on improved SSD, Yunnan Univ., 2022. https://doi.org/10.27456/d.cnki.gyndu.2022.001224
[25]	J. C. Liu, Deep learning-based cigarette appearance defect detection and classification, Yunnan Univ., 2022. https://doi.org/10.27456/d.cnki.gyndu.2022.002164
[26]	Y. Peng, Cigarette appearance quality detection based on improved YOLO deep learning model, Yunnan Univ. Fin. Econom., 2022. https://doi.org/10.27455/d.cnki.gycmc.2023.000862
[27]	Y. Peng, D. Jiang, X. Z. Lv, Y. Liu, Efficient and high-performance cigarette appearance detection based on YOLOv5, in 2023 International Conference on Intelligent Perception and Computer Vision (CIPCV), (2023), 7–12. https://doi.org/10.1109/CIPCV58883.2023.00010
[28]	X. M. Li, G. Q. Xie, Z. Huang, C. Yu, Cigarette appearance detection system based on cascaded convolution network, J. Comput. Appl., 43 (2023), 346–350. https://doi.org/10.11772/j.issn.1001-9081.2022030364 doi: 10.11772/j.issn.1001-9081.2022030364
[29]	M. Everingham, L. Van Gool, C.K.I. Williams, J. Winn, A. Zisserman, The PASCAL visual object classes (VOC) challenge, Int. J. Comput. Vis., 88 (2010), 303–338. https://doi.org/10.1007/s11263-009-0275-4 doi: 10.1007/s11263-009-0275-4
[30]	T. Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, et al., Microsoft coco: Common objects in context, in Computer Vision–ECCV 2014: 13th European Conference, 13 (2014), 740–755. https://doi.org/10.1007/978-3-319-10602-1_48
[31]	J. Deng, W. Dong, R. Socher, L. Li, K. Li, F. Li, ImageNet: A large-scale hierarchical image database, in 2009 IEEE Conference on Computer Vision and Pattern Recognition, (2009), 248–255. https://doi.org/10.1109/CVPR.2009.5206848
[32]	L. Zhang, B. W. Zhou, L. H. Wu, SSD network based on improved convolutional attention module and residual structure, Comput. Sci., 49 (2022), 211–217. http://qikan.cqvip.com/Qikan/Article/Detail?id=7106717136
[33]	J. Leng, Y. Liu, An enhanced SSD with feature fusion and visual reasoning for object detection, Neural Comput. Appl., 31 (2019), 6549–6558. https://doi.org/10.1007/s00521-018-3486-1 doi: 10.1007/s00521-018-3486-1
[34]	A. Graves, Practical variational inference for neural networks, Adv. Neural Inform. Process. Syst., 24 (2011), 2348–2356. https://dl.acm.org/doi/10.5555/2986459.2986721 doi: 10.5555/2986459.2986721
[35]	C. Blundell, J. Cornebise, K. Kavukcuoglu, D. Wierstra, Weight uncertainty in neural network, in Proceedings of the 32nd International Conference on Machine Learning, 37 (2015), 1613–1622. https://dl.acm.org/doi/10.5555/3045118.3045290
[36]	N. D. Nguyen, T. Do, T. D. Ngo, D. D. Le, An evaluation of deep learning methods for small object detection, J. Electr. Comput. Eng., (2020), 2348–2356. https://doi.org/10.1155/2020/3189691
[37]	Z. Zhu, D. Liang, S. Zhang, X. Huang, B. Li, S. Hu, Traffic-sign detection and classification in the wild, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2016), 2110–2118. https://doi.org/10.1109/CVPR.2016.232
[38]	A. Torralba, R. Fergus, W. T. Freeman, 80 million tiny images: A large data set for nonparametric object and scene recognition, IEEE Trans. Pattern Anal. Mach. Intell., 30 (2008), 1958–1970. https://doi.org/10.1109/TPAMI.2008.128 doi: 10.1109/TPAMI.2008.128
[39]	C. Chen, M. Y. Liu, O. Tuzel, et al., R-CNN for small object detection, in Asian Conference on Computer Vision(ACCV), 10115 (2014), 214–230. https://doi.org/10.1007/978-3-319-54193-8_14
[40]	S. Kullback, R. A. Leibler, On information and sufficiency, Annals Math. Stat., 22 (1951), 79–86.
[41]	D. P. Kingma, T. Salimans, M. Welling, Variational dropout and the local reparameterization trick, Adv. Neural Inform. Process. Syst., 2 (2015), 2575–2583. https://dl.acm.org/doi/abs/10.5555/2969442.2969527 doi: 10.5555/2969442.2969527
[42]	L. Tychsen-Smith, L. Petersson, Improving object localization with fitness NMS and bounded IoU loss, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2023), 6877–6885. https://doi.org/10.1109/CVPR.2018.00719
[43]	P.J. Huber, Robust estimation of a location parameter, Breakthr. Stat., (1992), 492–518. https://doi.org/10.1007/978-1-4612-4380-9_35
[44]	A. Krizhevsky, G. Hinton, Learning multiple layers of features from tiny images, Handbook Syst. Aut. Dis., 1 (2009).
[45]	K. He, G. Gkioxari, N. Parmar, P. Dollar, R. Girshick, Mask r-cnn, in Proceedings of the IEEE International Conference on Computer Vision (ICCV), (2017), 2961–2969. https://arXiv.org/abs/1703.06870
[46]	N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, S. Zagoruyko, End-to-end object detection with transformers, in European conference on computer vision (ECCV), (2020), 213–229. https://doi.org/10.1007/978-3-030-58452-8_13

Reader Comments

Your name:*

Email:*
© 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)