Evaluation of Data Augmentation for Image-Based Plant-Disease Detection

We discuss an evaluation of data augmentation to improve the accuracy in diagnosing plant diseases. Currently, research on image-based plant-disease detection using deep learning has been conducted and almost this type of research requires a huge amount of data but it is difficult to get them. Thus we choose applying data-augmentation to image-based plant-disease detection to solve this problem. In many cases, data-augmentation is effective; however, in some cases, it not be. Since the condition which data-augmentation deteriorates is not clear, further research is required. If train data and validation data are similar data, it is considered that diagnostic accuracy improves. Therefore, we measured the distance between training data and validation data by referring inception score and Frechet inception distance. Both scores are often used measuring the difference between two images. We adapted this idea to those data. If the distance is small, the diagnostic accuracy is improved, and vice versa. In this paper, we clarified the correlation between Frechet inception distance and diagnostic accuracy, inception score and diagnostic accuracy as indicators for determining whether it is suitable or not from the distance between train data and validation data expanded with five type data-augmentation.

[1]  Xavier Gastaldi,et al.  Shake-Shake regularization of 3-branch residual networks , 2017, ICLR.

[2]  Marcel Salathé,et al.  Using Deep Learning for Image-Based Plant Disease Detection , 2016, Front. Plant Sci..

[3]  Sepp Hochreiter,et al.  GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium , 2017, NIPS.

[4]  Huaiyu Zhu.  On Information and Sufficiency , 1997 .

[5]  Hitoshi Iyatomi,et al.  Basic Investigation on a Robust and Practical Plant Diagnostic System , 2016, 2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA).

[6]  Sang Cheol Kim,et al.  A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition , 2017, Sensors.

[7]  Andrea Luvisi,et al.  X-FIDO: An Effective Application for Detecting Olive Quick Decline Syndrome with Deep Learning and Data Fusion , 2017, Front. Plant Sci..

[8]  Hongyi Zhang,et al.  mixup: Beyond Empirical Risk Minimization , 2017, ICLR.

[9]  Svetlana Lazebnik,et al.  Multi-scale Orderless Pooling of Deep Convolutional Activation Features , 2014, ECCV.

[10]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[11]  Thomas Brox,et al.  Discriminative Unsupervised Feature Learning with Convolutional Neural Networks , 2014, NIPS.

[12]  Graham W. Taylor,et al.  Improved Regularization of Convolutional Neural Networks with Cutout , 2017, ArXiv.

[13]  Radhika Dave,et al.  Extreme vulnerability of smallholder farmers to agricultural risks and climate change in Madagascar , 2014, Philosophical Transactions of the Royal Society B: Biological Sciences.

[14]  Wojciech Zaremba,et al.  Improved Techniques for Training GANs , 2016, NIPS.

[15]  Sergey Ioffe,et al.  Rethinking the Inception Architecture for Computer Vision , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).