Farklı Çiçek Türlerini Derin Öğrenme Yöntemi İle Tanıma

Öz

Bu çalışmada derin öğrenme teknikleri kullanılarak çiçek türlerini tanıyabilen bir algoritma geliştirilmiştir. Bitkileri ve çiçekleri ayırt etmek araştırmacılar, ziraat mühendisleri, orman mühendisleri, çiftçiler ve botanikçiler için önem taşımaktadır. Çiçek türlerini ayırt edebilmek için her çiçeğe özgü özellik ve biçimlerin çıkarılması gerekmektedir. Çiçeklerin karmaşık arka planı, farklı çiçek türleri arasındaki benzerlik ve aynı çiçek türleri arasındaki farklılıklar nedeniyle çiçek görüntülerinin sınıflandırılması zorlu bir görevdir. Bu nedenle çiçek görüntülerinin bilgisayar ortamında tanınması ve gruplandırılması çeşitli kolaylıklar sağlayarak başarımı arttırmaktadır. Bu çalışmada bir çiçeğin görüntüsü sisteme yüklendiğinde çiçeğin türünü tahmin eden bir sistem geliştirilmiştir. Sistemin eğitiminde ve test işlemlerinde Oxford 102 veri seti kullanılmıştır. Bu veri seti 102 farklı çiçek türüne ait 7370 görüntü içermektedir. Çiçek türlerini sınıflandırmak için son yıllarda görüntü işleme konusundaki başarıları nedeniyle ResNet152 derin öğrenme mimarisi kullanılmıştır. Test görüntüleri için %99 sınıflama başarısı gösteren sistemin diğer çiçek sınıflama yöntemlerinden daha başarılı olduğu görülmüştür.

Anahtar Kelimeler

• Çiçek sınıflandırma
• derin öğrenme
• ResNet 152
• derin kalıntı ağlar

Recognizing The Different Flower Species By Deep Learning

Öz

In this study, an algorithm that can recognize flower species using deep learning techniques has been developed. Differentiating plants and flowers is important for researchers, agricultural engineers, forestry engineers, farmers and botanists. In order to distinguish flower types, features and forms specific to each flower need to be extracted. Classifying flower images is a challenging task because of the complex background of flowers, the similarity between different flower types, and the differences between the same flower types. For this reason, the recognition and grouping of flower images in computer environment increases the performance by providing various facilities. In this study, a system has been developed that predicts the type of flower when the image of a flower is loaded into the system. The Oxford 102 data set was used in the training and testing of the system. This data set contains 7370 images of 102 different flower species. Due to its achievements in image processing in recent years, ResNet152 deep learning architecture is used in this study to classify flower species. The system, which showed 99% classification success for test images, was found to be more successful than other flower classification methods.

Anahtar Kelimeler

• Flower Classification
• Deep Learning
• ResNet152
• Deep Residual Network

Kaynakça

1. Dinçer, D., Bekçi, B., & Bekiryazici, F. 2016. Türkiye’deki doğal bitki türlerinin üretiminde doku kültürü tekniklerinin kullanımı, Nevşehir Bilim ve Teknoloji Dergisi, Cilt. 5, s. 295-302. DOI: 10.17100/nevbiltek.211012
2. Desmond, R., & Ellwood, C. 2020. Dictionary of British and Irish botanists and horticulturists: including plant collectors, flower painters and garden designers. CRC Press, London, 900s.
3. Wu, Y., Qin, X., Pan, Y., & Yuan, C. 2018. Convolution neural network based transfer learning for classification of flowers. IEEE 3rd International Conference on Signal and Image Processing (ICSIP), July 13-15, 562-566.
4. Szegedy, C., Ioffe, S., Vanhoucke, V., & Alemi,, A. 2017. Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning, Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI'17), February 4-9, San Francisco, 4278–4284.
5. Jobin, A., Nair, M. S., Tatavarti, R. 2012. Plant Identification Based on Fractal Refinement Technique (FRT), Procedia Technology, Cilt. 6, s. 171-179. DOI: 10.1016/j.protcy.2012.10.021
6. Pauwels, E. J., Zeeuw, P. M., Ranguelova, E. B. 2009. Computer-Assisted Tree Taxonomy By Automated Image Recognition, Engineering Applications of Artificial Intelligence, Cilt. 22, s. 26-31. DOI: 10.1016/j.engappai.2008.04.017
7. Das, M., Manmatha, R. and Riseman, E.M. 1999. Indexing Flower Patent Images Using Domain Knowledge, IEEE Intelligent Systems and Their Applications, Cilt. 14, s. 24-33. DOI: 10.1109/5254.796084
8. Nilsback, M.A. and Zisserman, A. 2006. A Visual Vocabulary for Flower Classification, IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06), June 17-22, New York, USA, 1447-1454. DOI: 10.1109/CVPR.2006.42

9. Tseng, V., Wang, M., Su, J. 2005. A New Method for Image Classification by Using Multilevel Association Rules, 21st International Conference on Data Engineering Workshops (ICDEW'05), April 3-4, Tokyo, 1180-1188. DOI: 10.1109/ICDE.2005.164
10. Cho, S.-Y. and Lim, P.-H. 2006. A novel Virus Infection Clustering for Flower Images Identification, 18th International Conference on Pattern Recognition (ICPR'06), August 20-24, Hong Kong, 1038-1041. DOI: 10.1109/ICPR.2006.144
11. Nilsback, M. and Zisserman, A. 2008. Automated Flower Classification over a Large Number of Classes, Sixth Indian Conference on Computer Vision, Graphics & Image Processing, December 16-19, Bhubaneswar, 722-729. DOI: 10.1109/ICVGIP.2008.47
12. Liu, Y., Tang, F., Zhou, D., Meng Y. and Dong, W. 2016. Flower Classification Via Convolutional Neural Network, IEEE International Conference on Functional-Structural Plant Growth Modeling, Simulation, Visualization and Applications (FSPMA), November 7-11, Qingdao, 110-116. DOI: 10.1109/FSPMA.2016.7818296
13. Xia, X., Xu, C. and Nan, B. 2017. Inception-v3 for Flower Classification, 2nd International Conference on Image, Vision and Computing (ICIVC), June 2-4, Chengdu, 783-787. DOI: 10.1109/ICIVC.2017.7984661
14. Wu, Y., Qin, X., Pan, Y. and Yuan, C. 2018. Convolution Neural Network Based Transfer Learning for Classification of Flowers, IEEE 3rd International Conference on Signal and Image Processing (ICSIP), July 13-15, Shenzhen, 562-566. DOI: 10.1109/SIPROCESS.2018.8600536
15. Hiary, H., Saadeh, H., Saadeh, M., & Yaqub, M. 2018. Flower Classification Using Deep Convolutional Neural Networks. IET Computer Vision, 12(6), 855-862. DOI: 10.1049/iet-cvi.2017.0155
16. Deng, L. and Yu, D. 2013. Deep Learning: Methods and Applications, Foundations and Trends in Signal Processing, Cilt. 7, s. 197–387. DOI: 10.1561/2000000039
17. Bengio, Y. 2009. Learning Deep Architectures for AI, Foundations and Trends in Machine Learning, Cilt. 2, p. 1–127. DOI: 10.1561/2200000006
18. Khan, A., Sohail, A., Zahoora, U., & Qureshi, A. S. 2020. A survey of the recent architectures of deep convolutional neural networks. Artificial Intelligence Review, Cilt. 53, s. 5455-5516. DOI: 10.1007/s10462-020-09825-6
19. ResNet – Azure Machine Learning | Microsoft Docs. 2020. https://docs.microsoft.com/tr-tr/azure/machine-learning/algorithm-module-reference/resnet (Erişim Tarihi: 02.02.2021).
20. Amidi, A., Amidi, S. 2021. The Evaluation of Image Classification Explained 2021. https://stanford.edu/~shervine/blog/evolution-image-classification-explained (Erişim Tarihi: 02.02.2021).
21. Song H.A., Lee SY. 2013. Hierarchical Representation Using NMF. ss 466-473. Lee, M., Hirose, A., Hou, Z.G., Kil, R.M., ed. 2013. Neural Information Processing. ICONIP 2013. Lecture Notes in Computer Science, vol 8226, Springer, Berlin, Heidelberg, 638s. DOI: 10.1007/978-3-642-42054-2_58
22. Tsang, S.-H., 2018. Review: ResNet — Winner of ILSVRC 2015 (Image Classification, Localization, Detection).https://towardsdatascience.com/review-resnet-winner-of-ilsvrc-2015-image-classification-localization-detection-e39402bfa5d8 (Erişim Tarihi: 02.02.2021).
23. Danışman, T. 2020. Segmentation of Portrait Images Using a Deep Residual Network Architecture, Dokuz Eylül Üniversitesi Mühendislik Fakültesi Fen ve Mühendislik Dergisi, Cilt. 22, s. 569-580. DOI: 10.21205/deufmd.2020226523
24. Çarkacı , N. 2018. Derin Öğrenme Uygulamalarında En Sık kullanılan Hiper-parametreler. https://medium.com/deep-learning-turkiye/derin-ogrenme-uygulamalarinda-en-sik-kullanilan-hiper-parametreler-ece8e9125c4 (Erişim Tarihi: 02.02.2021).
25. Neto, S. 2019. Build An Image Classifier to Recognize 102 Different Species of Flowers. Towards Data Science. https://towardsdatascience.com/flower-species-classifier-c8893030cd90 (Erişim Tarihi: 02.02.2021).
26. Ting, K.M. Confusion Matrix. ss 209. Sammut, C., & Webb, G.I., ed. 2011. Encyclopedia of machine learning, Springer Science & Business Media, New York, 1031s.
27. Gogul, I., Kumar, V.S., 2017. Flower Species Recognition System Using Convolution Neural Networks and Transfer Learning, Fourth International Conference on Signal Processing, Communication and Networking (ICSCN), March 16-18, Chennai, 1-6. DOI: 10.1109/ICSCN.2017.8085675
28. Nguyen, T.T.N., Van Tuan Le, T.L.L., Vu, H., Pantuwong, N., Yagi, Y., 2016. Flower Species Identification Using Deep Convolutional Neural Networks, AUN/SEED-Net Regional Conference for Computer and Information Engineering, September 6-7, Kuala Lumpur, 1-6.