计算机科学 ›› 2020, Vol. 47 ›› Issue (7): 92-96.doi: 10.11896/jsjkx.190700093
谢源, 苗玉彬, 许凤麟, 张铭
XIE Yuan, MIAO Yu-bin, XU Feng-lin, ZHANG Ming
摘要: 注塑瓶表面缺陷检测是注塑成型工艺流程中的重要环节,但生产中存在缺陷的注塑瓶样本数量相对匮乏,使得应用深度学习算法进行缺陷检测时容易产生过拟合现象。针对上述问题,文中提出并构建一种半监督(Semi-supervised)深度卷积生成对抗网络(Deep Convolutional Generative Adversarial Network,DCGAN)模型。该模型首先使用HSV(Hue Saturation Va-lue)颜色空间转换与大津算法(Otsu)对原始注塑瓶图像进行预处理得到训练集;然后组合学习任务,使得DCGAN的无监督判别器与注塑瓶表面缺陷检测的监督分类器共享卷积层参数,同时修改损失函数,在DCGAN模型的Wasserstein距离中加入交叉熵;最后使用Adam优化器进行模型训练。实验结果表明,该模型能够准确分辨具有缺陷的注塑瓶样本,分类准确率达到98.65%。与传统的机器学习算法以及采用数据增强的卷积神经网络模型相比,所提模型的分类准确率更高,且较好地避免了过拟合现象,能满足注塑瓶生产中表面缺陷的自动检测需求。
中图分类号:
[1]GUO Z Y,LI D Q.Study on Deformation of Warping Fermentation starter of Injection Molded Products[J].Plastic Technology,2001(1):22-24. [2]REN G C,MIAO X Q,GUO Z G.Optimum design of the double-toggle clamping unit for injection molding machine[J].Machine Design and Manufacture,2009(1):18-19. [3]ZHANG H J,QUAN L,LI B.Comparative Study on Energy Ef-ficiency of the Electro-hydraulic Control System in Injection Molding Machine[J].Chinese Journal of Mechanical Enginee-ring,2012,48(8):180-187. [4]LV Z.Research on Key Techniques of Vision Inspection for Injection Molding Produets[D].Massachusetts:Northeastern University,2009. [5]ATHA D J,JAHANSHAHI M R.Evaluation of deep learning approaches based on convolutional neural networks for corrosion detection[J].Structural Health Monitoring,2018,17(5):1110-1128. [6]ZEILER M D,FERGUS R.Visualizing and understanding con-volutional networks[C]//European Conference on Computer Vision.Springer,Cham,2014:818-833. [7]BJERRUM E J.SMILES enumeration as data augmentation for neural network modeling of molecules[J].arXiv:1703.07076,2017. [8]CHEN W B,GUAN Z X,CHEN Y J.Data augmentation method based on conditional generative adversarial net model[J].Journal of Computer Applications,2018,38(11):3305-3311. [9]CHENG X Y,XIE L,ZHU J X,et al.Review of Generative Adversarial Network[J].Computer Science,2019,46(3):74-81. [10]ANTONIOU A,STORKEY A,EDWARDS H.Data Augmentation Generative Adversarial Networks[J].arXiv:1711.04340. [11]SHIN H C,TENENHOLTZ N A,ROGERS J K,et al.Medical Image Synthesis for Data Augmentation and Anonymization using Generative Adversarial Networks[J].arXiv:1807.10225. [12]RADFORD A,METZ L,CHINTALA S.Unsupervised repre-sentation learning with deep convolutional generative adversarial networks[J].arXiv:1511.06434,2015. [13]LI T,YING N,YU X,et al.Semi-supervised learning in unbalanced and heterogeneous networks[J].arXiv:1901.01696. [14]LASSERRE J A,BISHOP C M,MINKA T P.Principled Hybrids of Generative and Discriminative Models[C]// IEEE Computer Society Conference on Computer Vision & Pattern Recognition.2006. [15]SALAKHUTDINOV R,HINTON G E.Learning a Nonlinear Embedding by Preserving Class Neighbourhood Structure[J].Journal of Machine Learning Research,2007,2:412-419. [16]GULRAJANI I,AHMED F,ARJOVSKY M,et al.ImprovedTraining Of Wasserstein Gans[C]//Advances in Neural Information Processing Systems.2017:5767-5777. [17]GOODFELLOW I,BENGIO Y,COURVILLE A,et al.Deeplearning[M].Cambridge:MIT Press,2016. [18]KINGMA D P,BA J.Adam:A method for stochastic optimization[J].arXiv:1412.6980,2014. [19]MAO Y C,WANG J,CHEN X L,et al.Dam Defect Recognition and Classification Based on Feature Combination and CNN[J].Computer Science,2019,46(3):267-276. |
[1] | 武红鑫, 韩萌, 陈志强, 张喜龙, 李慕航. 监督和半监督学习下的多标签分类综述 Survey of Multi-label Classification Based on Supervised and Semi-supervised Learning 计算机科学, 2022, 49(8): 12-25. https://doi.org/10.11896/jsjkx.210700111 |
[2] | 苏丹宁, 曹桂涛, 王燕楠, 王宏, 任赫. 小样本雷达辐射源识别的深度学习方法综述 Survey of Deep Learning for Radar Emitter Identification Based on Small Sample 计算机科学, 2022, 49(7): 226-235. https://doi.org/10.11896/jsjkx.210600138 |
[3] | 侯夏晔, 陈海燕, 张兵, 袁立罡, 贾亦真. 一种基于支持向量机的主动度量学习算法 Active Metric Learning Based on Support Vector Machines 计算机科学, 2022, 49(6A): 113-118. https://doi.org/10.11896/jsjkx.210500034 |
[4] | 庞兴龙, 朱国胜. 基于半监督学习的网络流量分析研究 Survey of Network Traffic Analysis Based on Semi Supervised Learning 计算机科学, 2022, 49(6A): 544-554. https://doi.org/10.11896/jsjkx.210600131 |
[5] | 王宇飞, 陈文. 基于DECORATE集成学习与置信度评估的Tri-training算法 Tri-training Algorithm Based on DECORATE Ensemble Learning and Credibility Assessment 计算机科学, 2022, 49(6): 127-133. https://doi.org/10.11896/jsjkx.211100043 |
[6] | 李发光, 伊力哈木·亚尔买买提. 基于改进CenterNet的航拍绝缘子缺陷实时检测模型 Real-time Detection Model of Insulator Defect Based on Improved CenterNet 计算机科学, 2022, 49(5): 84-91. https://doi.org/10.11896/jsjkx.210400142 |
[7] | 彭云聪, 秦小林, 张力戈, 顾勇翔. 面向图像分类的小样本学习算法综述 Survey on Few-shot Learning Algorithms for Image Classification 计算机科学, 2022, 49(5): 1-9. https://doi.org/10.11896/jsjkx.210500128 |
[8] | 许华杰, 陈育, 杨洋, 秦远卓. 基于混合样本自动数据增强技术的半监督学习方法 Semi-supervised Learning Method Based on Automated Mixed Sample Data Augmentation Techniques 计算机科学, 2022, 49(3): 288-293. https://doi.org/10.11896/jsjkx.210100156 |
[9] | 侯宏旭, 孙硕, 乌尼尔. 蒙汉神经机器翻译研究综述 Survey of Mongolian-Chinese Neural Machine Translation 计算机科学, 2022, 49(1): 31-40. https://doi.org/10.11896/jsjkx.210900006 |
[10] | 方仲礼, 王喆, 迟子秋. 面向多标签小样本学习的双流重构网络 Dual-stream Reconstruction Network for Multi-label and Few-shot Learning 计算机科学, 2022, 49(1): 212-218. https://doi.org/10.11896/jsjkx.201100143 |
[11] | 刘立波, 苟婷婷. 融合深度典型相关分析和对抗学习的跨模态检索 Cross-modal Retrieval Combining Deep Canonical Correlation Analysis and Adversarial Learning 计算机科学, 2021, 48(9): 200-207. https://doi.org/10.11896/jsjkx.200600119 |
[12] | 赵敏, 刘惊雷. 基于高斯场和自适应图正则的半监督聚类 Semi-supervised Clustering Based on Gaussian Fields and Adaptive Graph Regularization 计算机科学, 2021, 48(7): 137-144. https://doi.org/10.11896/jsjkx.200800190 |
[13] | 王省, 康昭. 基于光滑表示的半监督分类算法 Smooth Representation-based Semi-supervised Classification 计算机科学, 2021, 48(3): 124-129. https://doi.org/10.11896/jsjkx.200700078 |
[14] | 储杰, 张正军, 汤鑫瑶, 黄振生. 基于加权样本和共识率的标记传播算法 Label Propagation Algorithm Based on Weighted Samples and Consensus-rate 计算机科学, 2021, 48(3): 214-219. https://doi.org/10.11896/jsjkx.191200103 |
[15] | 刘鑫, 黄沁元, 李强, 冉茂霞, 周颖, 杨天. 基于卷积神经网络和声振图像的磁瓦内部缺陷检测 Fault Detection for Arc Magnet Based on Convolutional Neural Network and Acoustic VibrationImage 计算机科学, 2021, 48(11A): 648-654. https://doi.org/10.11896/jsjkx.210100161 |
|