Convolutional Neural Network Pre-trained MobileNetV2 Untuk Deteksi Cacat Kain

• SUBAGJA PUTRA PRATAMA
• 14207002

---

ABSTRAK

ABSTRAK

Nama              : Subagja Putra Pratama

NIM                 : 14207002

Program Studi : Ilmu Komputer

Fakultas          : Teknologi Informasi

Jenjang           : Strata Dua (S2)

Konsentrasi     : Data Mining

Judul            : “Convolutional Neural Network Pre-trained MobileNetV2 Untuk Deteksi Cacat Kain”

Pada industri produksi kain, ketidakmampuan mendeteksi cacat pada kain dapat menempatkan perusahaan pada risiko kerugian finansial dan reputasi yang disebabkan oleh produk yang tidak bisa dikategorikan sebagai produk yang baik. Perkembangan teknologi yang semakin maju dan penggunaan teknik deteksi otomatis menjadi suatu solusi yang dapat diandalkan dan efektif untuk diterapkan. Dalam penelitian ini akan digunakan model Convolutional Neural Network (CNN) yang dibangun secara manual dan model pre-trained yaitu MobileNetV2. Kedua model berhasil diterapkan pada dataset TILDA-400 (64x64 patches) dengan arsitektur 3 lapisan konvolusi dengan kernel size 3x3. Pada setiap lapisan menggunakan aktivasi ReLU pada CNN manual dan arsitektur pada MobileNetV2mempunyai arsitektur dengan 17 Bottlenect Residual Block, dimana setiap blocknya menggunakan Depthwise Separable Convolusions dengan ukuran 3x3 dan menggunakan aktivasi ReLU6 pada MobileNetV2. Menggunakan Global Max Pooling sebelum dilanjutkan pada palisan Fully Connected dengan aktivasi Softmax. MobileNetV2 mendapatkan hasil cukup baik dengan mencapai nilai akurasi 77% pada data pengujian dan lebih tinggi daripada model CNN manual yang mencapai nilai akurasi 59% pada data pengujian.

KATA KUNCI

Cacat Kain,CNN,MobileNetV2

---

DAFTAR PUSTAKA

DAFTAR REFERENSI

[1]N.C.PavithranandB.K.P,“FabricDefectDetection?:AReview,”2023. [Online]. Available: https://ssrn.com/abstract=4523827.

[2]A.Saxena,“AnIntroductiontoConvolutionalNeuralNetworks,”Int. J. Res. Appl. Sci. Eng. Technol., vol. 10, no. 12, pp. 943–947, 2022, doi: 10.22214/ijraset.2022.47789.

[3]I.C.Education,“Whatareconvolutionalneuralnetworks(CNN)?,” Bdtechtalks.Com. 2020.

[4] T. Almeida, F. Moutinho, and J. P. Matos-Carvalho,“FabricDefectDetectionWithDeepLearningandFalseNegativeReduction,”IEEE Access, vol. 9, pp. 81936–81945, 2021, doi: 10.1109/ACCESS.2021.3086028.

[5]W.Ouyang,B.Xu,J.Hou,andX.Yuan,“FabricDefectDetectionUsingActivationLayerEmbeddedConvolutionalNeuralNetwork,”IEEE Access, vol. 7, no. 2, pp. 70130–70140, 2019, doi: 10.1109/ACCESS.2019.2913620.

[6]M.M.Khodier,S.M.Ahmed,andM.S.Sayed,“ComplexPattern Jacquard Fabrics Defect Detection Using Convolutional Neural Networks andMultispectralImaging,”IEEE Access, vol. 10, pp. 10653–10660, 2022, doi: 10.1109/ACCESS.2022.3144843.

[7]O.C.Study,“ClassificationofFabricDefects.”[Online].Available:https://www.onlineclothingstudy.com/2019/02/classification-of-fabricdefects.html#:~:text=A fabric defect corresponds to,faulty yarns or machine spoils.

[8]T.Sphere,“FabricDefects.”[Online].Available: https://www.textilesphere.com/2019/10/fabric-defects.html.

[9]T.Industry,“DifferenttypesoffabricDefectswithpictures.”[Online]. Available: https://www.textileindustry.net/different-types-of-fabric- 66 Program Studi Ilmu Komputer (S2) Universitas Nusa Mandiri defects/.

[10] C. Li et al.,“DataAugmentationforInertialSensorDatainCNNsfor CattleBehaviorClassification,”IEEE Sensors Lett., vol. 5, no. 11, pp. 1–4, 2021, doi: 10.1109/LSENS.2021.3119056.

[11]H.C.Park,I.P.Hong,S.Poudel,andC.Choi,“DataAugmentationBased onGenerativeAdversarialNetworksforEndoscopicImageClassification,” IEEE Access, vol. 11, no. April, pp. 49216–49225, 2023, doi: 10.1109/ACCESS.2023.3275173.

[12] D. Refai, S. Abo-Soud, and M. Abdel-Rahman,“DataAugmentationusing Transformers and Similarity Measures for Improving Arabic Text Classification,”IEEE Access, vol. 11, no. November, pp. 132516–132531, 2022, doi: 10.1109/ACCESS.2023.3336311.

[13]R.Takahashi,T.Matsubara,andK.Uehara,“DataAugmentationUsing Random Image Cropping and PatchingforDeepCNNs,”IEEE Trans. Circuits Syst. Video Technol., vol. 30, no. 9, pp. 2917–2931, 2020, doi: 10.1109/TCSVT.2019.2935128.

[14] D.Muchlinski,“Machinelearninganddeeplearning,”Elgar Encycl. Technol. Polit., pp. 114–118, 2022, doi: 10.4337/9781800374263.machine.learning.deep.

[15]S.Suganyadevi,V.Seethalakshmi,andK.Balasamy,“Areviewondeep learninginmedicalimageanalysis,”Int. J. Multimed. Inf. Retr., vol. 11, no. 1, pp. 19–38, 2022, doi: 10.1007/s13735-021-00218-1.

[16] M. Akay et al.,“DeepLearningClassificationofSystemicSclerosisSkin UsingtheMobileNetV2Model,”IEEE Open J. Eng. Med. Biol., vol. 2, pp. 104–110, 2021, doi: 10.1109/OJEMB.2021.3066097.

[17]Z.Li,F.Liu,W.Yang,S.Peng,andJ.Zhou,“ASurveyofConvolutionalNeuralNetworks:Analysis,Applications,andProspects,”IEEE Trans. Neural Networks Learn. Syst., vol. 33, no. 12, pp. 6999–7019, 2022, doi: 10.1109/TNNLS.2021.3084827. 67 Program Studi Ilmu Komputer (S2) Universitas Nusa Mandiri

[18] M. Momeny, A. M. Latif, M. Agha Sarram, R. Sheikhpour, and Y. D. Zhang, “Anoiserobustconvolutionalneuralnetworkforimage classification,”Results Eng., vol. 10, no. February, p. 100225, 2021, doi: 10.1016/j.rineng.2021.100225.

[19]A.W.Salehi,etal.,“AStudyofCNNandTransferLearninginMedical Imaging: Advantages,Challenges,FutureScope,”Sustain., vol. 15, no. 7, 2023, doi: 10.3390/su15075930.

[20]S.Kumaresan,K.S.J.Aultrin,S.S.Kumar,andM.D.Anand,“TransferLearningwithCNNforClassificationofWeldDefect,”IEEE Access, vol. 9, pp. 95097–95108, 2021, doi: 10.1109/ACCESS.2021.3093487.

[21]N.Nafi’iyah,C.Fatichah,D.Herumurti,E.R.Astuti,andR.H.Putra,“MobileNetV2EnsembleSegmentationforMandibularonPanoramic Radiography,”Int. J. Intell. Eng. Syst., vol. 16, no. 2, pp. 546–560, 2023, doi: 10.22266/ijies2023.0430.45.

[22]A.G.Howard,etal.,“MobileNets:EfficientConvolutionalNeural NetworksforMobileVisionApplications,”2017,[Online].Available: http://arxiv.org/abs/1704.04861.

[23] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L. C. Chen, “MobileNetV2:InvertedResidualsandLinearBottlenecks,”Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit., pp. 4510–4520, 2018, doi: 10.1109/CVPR.2018.00474.

[24]H.Wang,K.Li,andC.Xu,“ANewGenerationofResNetModelBased on Artificial Intelligence and Few Data Driven and Its Construction in Image RecognitionModel,”Comput. Intell. Neurosci., 2022, doi: 10.1155/2022/5976155.

[25]A.Sharma,“ConfusionMatrixinMachineLearning,” Www.Geeksforgeeks.Org, 2018.

[26]G.Zeng,“Ontheconfusionmatrixincreditscoringanditsanalytical properties,”Commun. Stat. - Theory Methods, vol. 49, no. 9, pp. 2080– 68 Program Studi Ilmu Komputer (S2) Universitas Nusa Mandiri 2093, 2020, doi: 10.1080/03610926.2019.1568485.

[27]Muthukrishnan,“UnderstandingtheClassificationreportthroughsklearn,” Medium.Com, 2018.

[28] S. J. Zahra, R. Sulaiman, S. M. M. Kahaki, and A. S. Prabuwono, “Harmonicmeanprojectionshapetransformforleafclassification,”J. Comput. Sci., vol. 16, no. 9, pp. 1212–1219, 2020, doi: 10.3844/JCSSP.2020.1212.1219.

[29]H.Xie,Y.Zhang,andZ.Wu,“FabricDefectDetectionMethodCombing ImagePyramidandDirectionTemplate,”IEEE Access, vol. 7, pp. 182320– 182334, 2019, doi: 10.1109/ACCESS.2019.2959880.

[30]Q.Liu,C.Wang,Y.Li,M.Gao,andJ.Li,“AFabricDefectDetection MethodBasedonDeepLearning,”IEEE Access, vol. 10, pp. 4284–4296, 2022, doi: 10.1109/ACCESS.2021.3140118.

[31]K.Dong,C.Zhou,Y.Ruan,andY.Li,“MobileNetV2ModelforImage Classification,”Proc. - 2020 2nd Int. Conf. Inf. Technol. Comput. Appl. ITCA 2020, pp. 476–480, 2020, doi: 10.1109/ITCA52113.2020.00106.

[32]Y.AufarandT.P.Kaloka,“Robustacoffeeleafdiseasesdetectionbased onMobileNetV2model,”Int. J. Electr. Comput. Eng., vol. 12, no. 6, pp. 6675–6683, 2022, doi: 10.11591/ijece.v12i6.pp6675-6683.

[33]Y.Gulzar,“FruitImageClassificationModelBasedonMobileNetV2with DeepTransferLearningTechnique,”Sustain., vol. 15, no. 3, 2023, doi: 10.3390/su15031906.

[34]Kaggle,“TILDA400(64X64patches).” https://www.kaggle.com/datasets/angelolmg/tilda-400-64x64-patches.