CONVOLUTIONAL NEURAL NETWORKS UNTUK KLASIFIKASI JENIS KENDARAAN DI INDONESIA

• EKO RAMANUDIN
• 14002442

---

ABSTRAK

 

ABSTRAK Eko Ramanudin 14002442
Ilmu Komputer
Teknologi Informasi Strata Dua (S2) Image Processing
“Convolutional Neural Networks Untuk Klasifikasi Jenis Kendaraan di Indonesia” Nama NIM
Program Studi Fakultas
Jenjang Konsentrasi Judul Kemacetan yang terjadi pada saat melakukan pembayaran jalan tol masih menjadi persoalan di Indonesia. Belum adanya sistem yang secara otomatis melakukan klasifikasi terhadap kendaraan yang memasuki gerbang adalah salah satu penyebabnya. Dengan berkembangnya teknologi, golongan kendaraan dapat dikenali secara otomatis menggunakan Neural Network. Salah satu metode terkini untuk klasifikasi citra yaitu menggunakan Convolutional Neural Network (CNN). Penelitian ini menggunakan 1.200 dataset citra kendaraan yang terdiri dari 5 kelas.
Dataset ini tergolong memiliki jumlah yang rendah untuk CNN, sehingga bisa menyebabkan model bekerja kurang maksimal karena berpotensi terjadi overfitting. Alhasil tingkat validasi akurasi akan turun jauh berbanding dengan training akurasi. Oleh karena itu beberapa teknik coba dilakukan pada penelitian ini dengan melakukan konfigurasi arsitektur CNN. Seperti penambahan jumlah layer konvolusi dan pooling, melakukan teknik regulasi batch normalization dan dropout, serta melakukan inisialisasi bobot. Setelah dilakukan testing pada data uji
menggunkan 5-fold cross validations, didapat hasil validitas akurasi terbaik mencapai 96.18%. Kata kunci: Convolutional Neural Network, CNN, Kendaraan, Klasifikasi Congestion that occurs when making toll road payments is still a problem in Indonesia. There is no system that automatically classifies vehicles that enter the
gate, which is one of the reasons. With the development oftechnology, the vehicle class can be recognized automatically using the Neural Network. One ofthe latest methodsfor image classification is using the Convolutional Neural Network (CNN). This study uses 1,200 vehicle image datasets consisting of5 classes. This data set
is a low numberfor CNN, so it can cause the model to work less than optimally due
to overfitting that occurs. So the validation level will drop considerably compared
to the accuracy training. Therefore, several techniques were tried in this study by configuring the CNN architecture. Such as increasing the number of convolution and pooling layers, performing batch normalization and dropout regulation techniques, and initializing weights. After testing the data, using 5-fold cross
validations, the results obtained the best accuracy validity reached 96.18%.
Keywords: Convolutional Neural Network, CNN, Vehicle, Classification
 

KATA KUNCI

Convolutional Neural Network

---

DAFTAR PUSTAKA

 

BAB 6
DAFTAR REFERENSI [1] E. R. I. (ERI) K. Polri, “JUMLAH DATA KENDARAAN,” 2022.
http://rc.korlantas.polri.go.id:8900/eri2017/laprekappolda.php [2] B. P. J. T. BPJT, “GOLONGAN JENIS KENDARAAN BERMOTOR
PADA JALAN TOL YANG SUDAH BEROPERASI,” 2007.
https://bpjt.pu.go.id/konten/golongan-kendaraan [3] N. Jahan, S. Islam, and M. F. A. Foysal, “Real-Time Vehicle Classification
Using CNN,” 2020 11th Int. Conf. Comput. Commun. Netw. Technol.
ICCCNT 2020, 2020, doi: 1 0.1109/ICCCNT49239.2020.9225623.
[4] L. Yao, “An Effective Vehicle Counting Approach Based on CNN,” 2019
IEEE 2nd Int. Conf. Electron. Commun. Eng. ICECE 2019, pp. 15-19,
2019, doi: 10.1 109/ICECE48499.2019.9058582. [5] N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R.
Salakhutdinov, “Dropout: A simple way to prevent neural networks from
overfitting,” J. Mach. Learn. Res., vol. 15, pp. 1929-1958, 2014.
[6] A. L. Pramana, E. Setyati, and Y. Kristian, “Model Cnn Lenet Dalam
Pengenalan Jenis Golongan Kendaraan Pada Jalan Tol,” J. Tek., vol. 12, no.
2, p. 65, 2020, doi: 10.30736/jt.vl3i2.469.
[7] A. Tri Nurolan, “Deteksi Dan Klasifikasi Jenis Kendaraan Berbasis
Pengolahan Citra Dengan Metode Convolutional Neural Network (Cnn),”
Deteksi Jenis Kendaraan, 2019.
[8] M. Liu, W. Wu, Z. Gu, Z. Yu, F. Qi, and Y. Li, “Deep learning based on
Batch Normalization for P300 signal detection,” Neurocomputing, vol. 275,
pp. 288-297, Jan. 2018, doi: 10.101 6/j.neucom.201 7.08.039.
[9] N. Sabri, “A Comparison between Average and Max-Pooling in
Convolutional Neural Network for Scoliosis Classification,” Int. J. Adv.
Trends Comput. Sci. Eng., vol. 9, no. 1.4, pp. 689-696, Sep. 2020, doi:
Program Studi Ilmu Komputer (S2) Nusa Mandiri
46 10.30534/ijatcse/2020/9791 .42020.
[10] S. Yun and A. Wong, “Where Should We Begin? A Low-Level
Exploration of Weight Initialization Impact on Quantized Behaviour of Deep Neural Networks,” J. Comput. Vis. Imaging Syst., vol. 6, no. 1, pp. 1- 5, 2021, doi: 10.15353/jcvis.v6il.3538.
[11] P. Thanapol, K. Lavangnananda, P. Bouvry, F. Pinel, and F. Leprevost, “Reducing Overfitting and Improving Generalization in Training Convolutional Neural Network (CNN) under Limited Sample Sizes in
Image Recognition,” in 2020 - 5th International Conference on Information
Technology (InCIT), Oct. 2020, pp. 300-305. doi: 10.1 109/InCIT50588.2020.9310787.
[12] A. Rofiq Mulyana, A. Bejo, and H. Adi Nugroho, “The Improvement of
Character Recognition on ANPR Algorithm using CNN Method with Batch Normalization,” in 2021 International Seminar on Machine Learning, Optimization, and Data Science (ISMODE), Jan. 2022, pp. 168-172. doi:
10. 1 1 09/ISMODE53584.2022.9742804.
[13] J. Cai, O. Chang, X.-L. Tang, C. Xue, and C. Wei, “Facial Expression Recognition Method Based on Sparse Batch Normalization CNN,” in 2018 37th Chinese Control Conference (CCC), Jul. 2018, pp. 9608-9613. doi: 10.2391 9/ChiCC.2018.8483567.
[14] K. Sanjar, A. Rehman, A. Paul, and K. JeongHong, “Weight Dropout for Preventing Neural Networks from Overfitting,” in 2020 8th International
Conference on Orange Technology (ICOT), Dec. 2020, pp. 1-4. doi: 10.1 109/ICOT51877.2020.9468799.
[15] X. Ben and K. Jiang, “Image Classification of Damage of Houses hit by the hurricane based on Convolution Neural Network,” in 2021 International
Conference on Big Data, Artificial Intelligence and Risk Management (ICBAR), Nov. 2021, pp. 79-82. doi: 10.1 109/ICBAR55169.2021.00026.
[16] Ian J. Goodfellow and Yoshua Bengio and Aaron Courville, Deep Learning. MIT Press, 2016.
Program Studi Ilmu Komputer (S2) Nusa Mandiri
47
[17] M. Z. Alom, “Improved Deep Convolutional Neural Networks (DCNN)
Approaches for Computer Vision and Bio-Medical Imaging,” University of Dayton, 2018.
[18] Y. Xiao, “Vehicle Detection in Deep Learning,” 2019, [Online]. Available:
http://arxiv.org/abs/1905.13390
[19] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” in Advances in Neural
Information Processing Systems, 2012, vol. 25. [Online]. Available:
https://proceedings.neurips.cc/paper/2012/file/c399862d3b9d6b76c8436e9 24a68c45b-Paper.pdf
[20] W. Hu, Y. Huang, L. Wei, F. Zhang, and H. Li, “Deep Convolutional
Neural Networks for Hyperspectral Image Classification,” J. Sensors, vol. 2015, pp. 1-12, 2015, doi: 10.1155/2015/258619.
[21 ] Matlab., “2-D Convolutional Layer,” Retrievedfrom Mathworks.
https://www.mathworks.com/help/deeplearning/ref/nnet.cnn.layer.convolut ion2dlayer.html;jsessionid=l 1 3392aaf58692da9f0c47dc732e
[22] V. Bui and L. C. Chang, “Deep learning architectures for hard character
classification,” Proc. 2016 Int. Conf. Artif. Intell. ICAI 2016 -
WORLDCOMP 2016, pp. 108-1 14, 2016.
[23] M. Sarigul, B. M. Ozyildirim, and M. Avci, “Differential convolutional
neural network,” Neural Networks, vol. 1 16, pp. 279-287, Aug. 2019, doi: 10.1016/j.neunet.2019.04.025.
[24] P. Refaeilzadeh, L. Tang, and H. Liu, “Cross-Validation,” in Encyclopedia
ofDatabase Systems, Boston, MA: Springer US, 2009, pp. 532-538. doi:
10. 1 007/978-0-387-39940-9_565.
[25] S. Manna, “K-Fold Cross Validation for Deep Learning Models using Keras,” Medium, 2020. https://medium.com/the-owl/k-fold-crossvalidation-in-keras-3ec4a3a00538
[26] S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network
Program Studi Ilmu Komputer (S2) Nusa Mandiri
48 training by reducing internal covariate shift,” 32nd Int. Conf. Mach. Learn.
ICML 2015, vol. 1, pp. 448-456, 2015.
[27] K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image
Recognition,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jun. 2016, pp. 770-778. doi: 10.1 109/CVPR.2016.90.
[28] X. Glorot and Y. Bengio, “Understanding the difficulty of training deep feedforward neural networks,”./. Mach. Learn. Res., vol. 9, pp. 249-256, 2010.
[29] I. Markoulidakis, I. Rallis, I. Georgoulas, G. Kopsiaftis, A. Doulamis, and N. Doulamis, “Multiclass Confusion Matrix Reduction Method and Its
Application on Net Promoter Score Classification Problem,” Technologies,
vol. 9, no. 4, p. 81, Nov. 2021, doi: 10.3390/technologies9040081.
[30] A. T. Sasongko and M. Ivan Fanany, “Indonesia Toll Road Vehicle
Classification Using Transfer Learning with Pre-trained Resnet Models,” in
2019 International Seminar on Research ofInformation Technology and Intelligent Systems (ISR1TI), Dec. 2019, pp. 373-378. doi: 10.11 09/ISRITI48646.201 9.9034590.