Using Tensor Processing Units to enhance the training of Convolutional Neural Networks in Multiclass Classification

K.V.N. Rajesh¹ , D. Lalitha Bhaskari²

Section:Research Paper, Product Type: Journal-Paper
Vol.9 , Issue.1 , pp.48-55, Feb-2021

Online published on Feb 28, 2021

Copyright © K.V.N. Rajesh, D. Lalitha Bhaskari . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
 

View this paper at   Google Scholar | DPI Digital Library

XML View     PDF Download

How to Cite this Paper

Abstract :
Ayurveda is the ancient medical science of India. Many of the Ayurvedic medicines are plant-based. There is growing interest in Ayurveda nowadays and hence there is a need to make it more relevant to the current era by using modern technologies in Ayurveda. Identification of required Ayurvedic plants among a host of other plants is a problem that can be solved using computer vision-based technology. In this research work, classification of plants is done using Convolution Neural Networks (CNN) on the images and labels of respective leaves. A leaf image dataset for seven different Ayurvedic plants has been created as a part of this research work. Keras deep learning framework with TensorFlow as the backend is used for building the CNN model. The platform used for training this model was Google Colab which is available free of cost. As a part of this research work, Method and program are developed to load data from normal storage like Google drive, to TPU. After loading the image data to TPU, the TPU hardware acceleration is used for training the CNN model. In addition to a custom CNN model developed from scratch, pretrained image recognition models like DenseNet201, EfficientNetB7, InceptionV3, ResNet50V2, VGG19 and Xception are leveraged as a part of this research work. Image recognition accuracies ranging from 95% to 100% have been achieved using the mentioned CNN training methods for plant leaf image classification.

Key-Words / Index Term :
Ayurvedic Plant Leaf Images, Custom Leaf Image Dataset, Computer Vision, Multiclass Classification, Convolution Neural Networks, Pre-trained image recognition models, Tensor Processing Unit

References :
[1] Luna, R.G.d., Baldovino, R.G., Cotoco, E.A., Ocampo, A.L.P.d., Valenzuela, I.C., Culaba, A.B. and Dadios, E.P. “Identification of Philippine Herbal Medicine Plant Leaf Using Artificial Neural Network,” In the proceedings of 2017 IEEE 9th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment and Management (HNICEM), Manila. pp. 1-8,2017.
[2] Liu, W., Li, Z., Lin, J. and Liang, D. “The PSO-SVM-based Method of the Recognition of Plant Leaves,” In the proceedings of 2017 IEEE International Symposium on Parallel and Distributed Processing with Applications and 2017 IEEE International Conference on Ubiquitous Computing and Communications (ISPA/IUCC). pp. 1350-1355. 2017.
[3] Beikmohammadi, A. and Faez, K., “Leaf Classification for Plant Recognition with Deep Transfer Learning,” In the proceedings of the 2018 4th Iranian Conference on Signal Processing and Intelligent Systems (ICSPIS,. pp. 21-26,2018.
[4] Liu, J., Yang, S., Cheng, Y. and Song, Z. “Plant Leaf Classification Based on Deep Learning,” 2018 Chinese Automation Congress (CAC). pp. 3165-3169. 2018.
[5] Liantoni, F., Perwira, R.I., Muharom, S., Firmansyah, R.A. and Fahruzi, A. “Leaf classification with improved image feature based on the seven moment invariant,” Journal of Physics: Conference Series, In the proceedings of 1st International Conference on Advance and Scientific Innovation , Medan, Indonesia IOP Publishing.23–24 April 2018.
[6] Kayhan, G. and Ergün, E.”Medicinal and Aromatic Plants Identification Using Machine Learning Methods.” Balkan Journal of Electrical & Computer Engineering, 8(1), pp. 81-87, 2020.
[7] Chompookham, T., Gonwirat, S., Lata, S. and Phiphiphatphaisit, S. “Plant Leaf Image Recognition using Multiple grid Based Local Descriptor and Dimensionality Reduction Approach,” ICISS 2020,In the Proceedings of the 2020 The 3rd International Conference on Information Science and System. pp. 72-77,2020.
[8] Tang, Z. (2020) “Leaf image recognition and classification based on GBDT-probabilistic neural network,”Journal of Physics: Conference Series, In the Third International Conference on Physics, Mathematics and Statistics (ICPMS2020), Kunming, China IOP Publishing Ltd, May 20-22, 2020.
[9] Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke and Andrew Rabinovich, “Going deeper with convolutions,” In the Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),Boston,MA. pp. 1-9,2015.
[10] Karen Simonyan and Andrew Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition.”In the proceedings of International Conference on Learning Representations, 2015.
[11] Kaiming He, Xiangyu Zhang, Shaoqing Ren and Jian Sun, “Deep Residual Learning for Image Recognition,”In the proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition”,Las Vegas, NV, pp. 770-778,2016.
[12] F.Chollet “Xception: Deep Learning with Depthwise Separable Convolution,” In the proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, pp. 1800-1807.2017.
[13] G. Huang, Z. Liu, L. Van Der Maaten and K. Q. Weinberger, “Densely Connected Convolutional Networks,”In the proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),Honolulu,HI, pp. 2261-2269,2017.
[14] Tan, M. and Quoc V. Le EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. ICML.2019.