Ahşap Kusur Tespiti İçin Optimize Edilmiş AlexNet Mimarisi

Abstract

Bu makale, AlexNet mimarisi kullanılarak kusurlu ve kusursuz ahşap yüzey görüntülerinin sınıflandırılması üzerine odaklanmaktadır. İlk olarak, karışık olan yüzey görüntüleri kusurlu ve kusursuz diye ikiye ayrılmış ve yeniden düzenlenmiştir. Bu veri kümesinde 1992 kusursuz, 18 284 kusurlu ahşap yüzey görüntüsü bulunmaktadır. Ahşap yüzey görüntüleri üzerinde toplam 43 000 ahşap kusur bulunmaktadır. AlexNet mimarisi transfer öğrenme yaklaşımı kullanılarak deneyler gerçekleştirilmiştir. Deneylerde, farklı epoch sayıları (25 epoch, 50 epoch) ve veri artırma yöntemi kullanılarak AlexNet modelin eğitimi gerçekleştirilmiş ve sonra test edilmiştir. Ahşap yüzey kusur tespitinde ikili sınıflamada sonuç olarak, AlexNet mimarisi ile kusurlu ve kusursuz ahşap yüzey görüntülerinin sınıflandırılması sonucunda en başarılı sonuçları AlexNet Augmented* modelinin 50 epoch sonrasında elde ettiği görülmektedir. Bu modelde, doğruluk değeri 0.9687, AUC değeri 0.9892 olarak hesaplanmıştır. Yaklaşık %97 oranında ahşap kusur tespiti sonucu bu çalışmada elde edilmiştir. Ayrıca, hassasiyet, geri çağırma ve F-skor değerleri de 0.97 olarak belirlenmiştir. Bu sonuçlar, ahşap yüzey kusur tespitinde AlexNet modelinin yüksek bir performans sergilediğini göstermektedir.

Keywords

• Ahşap kusur
• Bilgisayarlı görü
• Yapay zeka
• Derin öğrenme
• Transfer öğrenme

AlexNet Architecture Optimized for Wood Defect Detection

Abstract

This paper focuses on the classification of imperfect and perfect wood surface images using AlexNet architecture. Firstly, the mixed surface images are divided into imperfect and perfect and reorganised. This dataset contains 1992 undefective and 18 284 defective wood surface images. There are a total of 43 000 wood defects on this dataset. Experiments are carried out using the AlexNet architecture transfer learning approach. In the experiments, the AlexNet model is trained using different epoch numbers (25 epochs, 50 epochs) and data augmentation method. It is then tested. As a result of binary classification in wood surface defect detection, it is seen that the AlexNet Augmented* model obtained the most successful results after 50 epochs as a result of the classification of defective and perfect wood surface images with AlexNet architecture. In this model, the accuracy rate is calculated as 0.9687 and AUC value as 0.9892. Approximately 97% of wood defect detection results are obtained in this study. In addition, the precision, recall and F-score values are determined as 0.97. These results show that the AlexNet model has a high performance in wood surface defect detection.

Keywords

• Wood defect
• Computer vision
• Artifical intelligence
• Deep learning
• Transfer learning

References

1. [1] S. Lee, S. J. Lee, J. S. Lee, K. B. Kim, J. J. Lee, and H. Yeo, "Basic study on nondestructive evaluation of artificial deterioration of a wooden rafter by ultrasonic measurement," Journal of Wood Science, vol. 57, pp. 387-394, 2011, doi: 10.1007/s10086-011-1186-x.
2. [2] H. Xu, L. Wang, and S. Ni, "Application of Artificial Neural Network to Nondestructive Testing of Internal Wood Defects Based on the Intrinsic Frequencies," in 2010 International Conference on System Science, Engineering Design and Manufacturing Informatization, vol. 1, pp. 207-210, Nov. 2010, doi: 10.1109/ICSEM.2010.63.
3. [3] Z.F. Qiu, "A Simple Machine Vision System for Improving the Edging and Trimming Operations Performed in Hardwood Sawmills," Master's Thesis, Virginia Polytechnic Institute and State University, Blacksburg, VA, USA, 1996.
4. [4] D.L. Schmoldt, P. Li, and A.L. Abbott, "Machine vision using artificial neural networks with local 3D neighborhoods," Computers and Electronics in Agriculture, vol. 16, pp. 255-271, 1997, doi: 10.1016/S0168-1699(97)00002-1.
5. [5] D.W. Qi, P. Zhang, X. Jin, and X. Zhang, "Study on wood image edge detection based on Hopfield neural network," in Proceedings of the 2010 IEEE International Conference on Information and Automation, Harbin, China, 20-23 June 2010, pp. 1942-1946, doi: 10.1109/ICINFA.2010.5512014.
6. [6] X.Y. Ji, H. Guo, and M.H. Hu, "Features Extraction and Classification of Wood Defect Based on Hu Invariant Moment and Wavelet Moment and BP Neural Network," in Proceedings of the 12th International Symposium on Visual Information Communication and Interaction (VINCI'2019), Shanghai, China, 20-22 September 2019, Article 37, pp. 1-5, Association for Computing Machinery: New York, NY, USA, 2019, doi: 10.1145/3356422.3356459.
7. [7] H. Mu and D.W. Qi, "Pattern Recognition of Wood Defects Types Based on Hu Invariant Moments," in Proceedings of the 2009 2nd International Congress on Image and Signal Processing, Tianjin, China, 17-19 October 2009, pp. 1-5, doi : 10.1109/CISP.2009.5303866.
8. [8] J.C. Hermanson and A.C. Wiedenhoeft, "A brief review of machine vision in the context of automated wood identification systems," IAWA Journal, vol. 32, pp. 233-250, 2011.

9. [9] L. Wen, X.Y. Li, and L. Gao, "A transfer convolutional neural network for fault diagnosis based on ResNet-50," Neural Computing and Applications, vol. 32, pp. 6111-6124, 2020, doi: 10.1007/s00521-019-04097-w.
10. [10] LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444, doi:10.1038/nature14539.
11. [11] R. Zhao, R.Q. Yan, Z.H. Chen, K.Z. Mao, P. Wang, and R.X. Gao, "Deep learning and its applications to machine health monitoring," Mechanical Systems and Signal Processing, vol. 115, pp. 213-237, 2019, doi: 10.1016/j.ymssp.2018.05.050.
12. [12] J. Donahue, Y.Q. Jia, O. Vinyals, J. Hoffman, N. Zhang, E. Tzeng, and T. Darrell, "Decaf: A deep convolutional activation feature for generic visual recognition," in Proceedings of the International Conference on Machine Learning, Beijing, China, 21-26 June 2014, pp. 647-655.
13. [13] J. Yosinski, J. Clune, Y. Bengio, and H. Lipson, "How transferable are features in deep neural networks?" in Proceedings of the Advances in Neural Information Processing Systems, Montreal, QC, Canada, 8-13 December 2014, pp. 3320-3328.
14. [14] K. Thenmozhi and U.S. Reddy, "Crop pest classification based on deep convolutional neural network and transfer learning," Computers and Electronics in Agriculture, vol. 164, article 104906, 2019, doi: 10.1016/j.compag.2019.104906.
15. [15] X. Gao, Y.F. Zhao, Q. Xiong, and Z. Chen, "Identification of Tree Species Based on Transfer Learning," Forest Engineering, vol. 35, pp. 68-75, 2019.
16. [16] S. Kentsch, M.L. Lopez Caceres, D. Serrano, F. Roure, and Y. Diez, "Computer Vision and Deep Learning Techniques for the Analysis of Drone-Acquired Forest Images, a Transfer Learning Study," Remote Sensing, vol. 12, no. 8, article 1287, 2020, doi: 10.3390/rs12081287.
17. [17] P. Kodytek, A. Bodzas, and P. Bilik, "A large-scale image dataset of wood surface defects for automated vision-based quality control processes [version 2]," F1000Research, vol. 10, article 581, 2022, doi: 10.12688/f1000research.52903.2
18. [18] J. Gu, Z. Wang, J. Kuen, L. Ma, A. Shahroudy, B. Shuai, and T. Chen, "Recent advances in convolutional neural networks," Pattern Recognition, vol. 77, pp. 354-377, 2018, doi: 10.1016/j.patcog.2017.10.013.
19. [19] N. Erbaş, G. Çınarer, and K. Kılıç, "Classification of hazelnuts according to their quality using deep learning algorithms," Czech Journal of Food Sciences, vol. 40, no. 3, pp. 240-248, 2022, doi: 10.17221/21/2022-CJFS.
20. [20] E. Ergün and K. Kılıç, "Derin öğrenme ile artırılmış görüntü seti üzerinden cilt kanseri tespiti," Black Sea Journal of Engineering and Science, vol. 4, no. 4, pp. 192-200, 2021, doi: 10.34248/bsengineering.938520.
21. [21] A. Krizhevsky, I. Sutskever, and G.E. Hinton, "ImageNet classification with deep convolutional neural networks," in Advances in Neural Information Processing Systems, pp. 1097-1105, 2012, doi: 10.1145/3065386.
22. [22] J. Yosinski, J. Clune, Y. Bengio, and H. Lipson, "How transferable are features in deep neural networks?" in Advances in Neural Information Processing Systems, pp. 3320-3328, 2014.
23. [23] R. Ren, T. Hung, and K. C. Tan, "A generic deep-learning-based approach for automated surface inspection," IEEE Transactions on Cybernetics, vol. 48, no. 3, pp. 929-940, 2017 doi: 10.1109/TCYB.2017.2668395.
24. [24] Y. Zhang, C. Xu, C. Li, H. Yu, and J. Cao, "Wood defect detection method with PCA feature fusion and compressed sensing," Journal of Forestry Research, vol. 26, pp. 745-751, 2015, doi: 10.1007/s11676-015-0066-4.
25. [25] H. Yu, Y. Liang, H. Liang, and Y. Zhang, "Recognition of wood surface defects with near infrared spectroscopy and machine vision," Journal of Forestry Research, vol. 30, no. 6, pp. 2379-2386, 2019, doi: 10.1007/s11676-018-00874-w.
26. [26] Y. X. Zhang, Y. Q. Zhao, Y. Liu, L. Q. Jiang, and Z. W. Chen, "Identification of wood defects based on LBP features," in 2016 35th Chinese Control Conference (CCC), pp. 4202-4205, July 2016, doi: 10.1109/ChiCC.2016.7554010.