植物病蟲害辨識是科技結合農業領域中的一個重要主題，因為植物病蟲害辨識可以節省人力成本並且幫助農業發展。在目前的植物病蟲害辨識相關研究中，許多論文將主流CNN模型應用於辨識植物的疾病，這些研究透過葉子上的病徵來判斷該植物感染的疾病，然而大部分研究假設一片葉子上僅存在一種病徵，因此所採用的資料集皆為single-label data。但在實際的農園現場中，一株植物可能同時間會感染上多種疾病，其葉片上也會出現多種病徵，因此本論文提出一個基於multi-label的植物病蟲害辨識系統來解決此問題。
本論文提出一個基於multi-task learning架構的multi-label分類模型，其為Multi-Label Branch Network (MLBNet)。Multi-task learning的概念是將多個任務整合在一個模型中，其中包含多個負責訓練自己的任務的分支架構，並且每個分支可以透過共享特徵來提高模型的泛化能力。在本系統中我們將multi-label分類問題轉換為多個single-label的二元分類問題，我們建立了多個分支架構來處理多個single-label二元分類問題，其中每個分支都可以被視為一個任務。在MLBNet中每個分支共享相同feature maps，並負責辨識其對應的類別，每個分支的輸出代表一張圖片中的葉子是否感染了該疾病。我們在multi-label dataset和single-label dataset上評估我們提出的方法。在multi-label的實驗中，我們將MLBNet和其他multi-label模型互相比較；實驗結果證明相較於其他multi-label方法相比我們提出的方法具有更好的準確率。而在single-label實驗中，我們評估MLBNet的表現並與多個著名的CNN模型進行比較；實驗結果顯示MLBNet在single-label分類中可以達到與效果最好的CNN模型相同的準確度。

Plant diseases and pests recognition is an important topic of agriculture technologies since it can reduce human labor and help plant growth. Many studies adopt deep learning models to identify plant diseases through investigating the symptoms on leaves. However, most approaches focus on the single-label problem that one left only has a single symptom. In practical applications, plants are possible to be infected with multiple diseases so that multiple symptoms could appear on the same leaf. The thesis developed a recognition system for plant diseases and pests based on multi-label learning to face the practical problem.
The study proposed a multi-label classification model, Multi-Label Branch Network (MLBNet), based on multi-task learning. The concept of multi-task learning is to integrate many tasks into a model, which contains multiple branch structures responsible for training its tasks. Each branch can share features to improve the model’s generalization ability. A multi-label classification problem can be transformed into several single-label binary classification problems. We construct multiple branches to deal with these single-label binary classifications, where each branch is regarded as an independent task. In our MLBNet, each branch shares the same feature map from previous layers and is trained for identifying the corresponding label. We evaluate our MLBNet on multi-label datasets and single-label datasets. In the experiments of multi-label classification, we also compared our proposed method with state-of-the-art methods. The results show that our proposed method outperforms other multi-label methods. And in single-label experiments, we evaluate MLBNet and then compare it with other popular deep CNN models like ResNet and DenseNet. The results show that MLBNet can achieve competitive accuracy compared to the best accuracy in deep CNN models.

Outline i
List of Figures ii
List of Tables iii
1. Introduction 1
2. Related work 3
2.1 Image Classification 3
2.2 Plant Diseases and Pests Recognition 4
2.3 Multi-label Learning 5
2.4 Multi-task Learning 6
3. Method 7
3.1 Proposed Method 7
3.2 Backbone Network 9
3.3 Multiple Branches Classifier 10
3.4 Network Design 11
3.5 Loss Function 14
3.6 Datasets 14
3.6.1 Pomelo Dataset 14
3.6.2 PlantVillage Dataset 19
4. Experimental Results 22
4.1 Experimental Setup 22
4.2 Evaluation Metric 22
4.3 Multi-label Classification 23
4.4 Single-label Classification Results 25
4.4.1 Pomelo-SL 25
4.4.2 PlantVillage 26
4.5 Discussion 27
4.5.1 Multi-label and single-label classification 28
4.5.2 Parameters 33
4.5.3 Learning Curves 33
4.5.4 Prediction results 34
5. Conclusion 37
6. Reference 38

[1]M. Everingham, L. Gool, C. K. Williams, J. Winn, and A. Zisserman, "The Pascal Visual Object Classes (Voc) Challenge," International Journal of Computer Vision, vol. 88, pp. 303-338, 2009.
[2]O. Russakovsky et al., "Imagenet Large Scale Visual Recognition Challenge," International Journal of Computer Vision, vol. 115, no. 3, pp. 211-252, 2015.
[3]J. Deng, W. Dong, R. Socher, L. Li, L. Kai, and F.-F. Li, "Imagenet: A Large-Scale Hierarchical Image Database," in 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp. 248-255, 2009.
[4]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.
[5]S. P. Mohanty, D. P. Hughes, and M. Salathé, "Using Deep Learning for Image-Based Plant Disease Detection," Frontiers in Plant Science, vol. 7, no. 1419, 2016-September-22 2016.
[6]M. Brahimi, K. Boukhalfa, and A. J. A. A. I. Moussaoui, "Deep Learning for Tomato Diseases: Classification and Symptoms Visualization," Applied Artificial Intelligence, vol. 31, no. 4, pp. 299-315, 2017.
[7]K. P. J. C. Ferentinos and E. i. Agriculture, "Deep Learning Models for Plant Disease Detection and Diagnosis," Computers and Electronics in Agriculture, vol. 145, pp. 311-318, 2018.
[8]A. Fuentes, S. Yoon, S. C. Kim, and D. S. Park, "A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition," Sensors, vol. 17, no. 9, p. 2022, 2017.
[9]M. TÜRKOĞLU, D. J. T. J. o. E. E. Hanbay, and C. Sciences, "Plant Disease and Pest Detection Using Deep Learning-Based Features," Turkish Journal of Electrical Engineering & Computer Sciences, vol. 27, no. 3, pp. 1636-1651, 2019.
[10]K. Zhang, Q. Wu, A. Liu, and X. J. A. i. M. Meng, "Can Deep Learning Identify Tomato Leaf Disease?," in Advances in Multimedia, 2018, vol. 2018.
[11]E. C. Too, L. Yujian, S. Njuki, and L. Yingchun, "A Comparative Study of Fine-Tuning Deep Learning Models for Plant Disease Identification," Computers and Electronics in Agriculture, vol. 161, pp. 272-279, 2019.
[12]Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, "Gradient-Based Learning Applied to Document Recognition," Proceedings of the IEEE, vol. 86, no. 11, pp. 2278-2324, 1998.
[13]K. Simonyan and A. Zisserman, "Very Deep Convolutional Networks for Large-Scale Image Recognition," in The International Conference on Learning Representations, 2015.
[14]C. Szegedy et al., "Going Deeper with Convolutions," in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1-9, 2015.
[15]K. He and J. Sun, "Convolutional Neural Networks at Constrained Time Cost," in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 5353-5360, 2015.
[16]K. He, X. Zhang, S. Ren, and J. Sun, "Deep Residual Learning for Image Recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770-778, 2016.
[17]G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, "Densely Connected Convolutional Networks," in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 4700-4708, 2017.
[18]D. Hughes and M. Salathé, "An Open Access Repository of Images on Plant Health to Enable the Development of Mobile Disease Diagnostics," arXiv preprint arXiv:1511.08060, 2015.
[19]M. R. Howlader, U. Habiba, R. H. Faisal, and M. M. Rahman, "Automatic Recognition of Guava Leaf Diseases Using Deep Convolution Neural Network," in 2019 International Conference on Electrical, Computer and Communication Engineering, pp. 1-5: IEEE, 2019.
[20]B. A. Ashqar and S. S. Abu-Naser, "Image-Based Tomato Leaves Diseases Detection Using Deep Learning," International Journal of Academic Engineering Research, vol. 2, no. 12, pp. 10-16, 2018.
[21]G. Wang, Y. Sun, and J. Wang, "Automatic Image-Based Plant Disease Severity Estimation Using Deep Learning," Computational intelligence and neuroscience, vol. 2017, 2017.
[22]E. Suryawati, R. Sustika, R. S. Yuwana, A. Subekti, and H. F. Pardede, "Deep Structured Convolutional Neural Network for Tomato Diseases Detection," in 2018 International Conference on Advanced Computer Science and Information Systems, pp. 385-390: IEEE, 2018.
[23]J. Amara, B. Bouaziz, and A. Algergawy, "A Deep Learning-Based Approach for Banana Leaf Diseases Classification," in Datenbanksysteme für Business, Technologie und Web-Workshopband, 2017.
[24]A. Ramcharan, K. Baranowski, P. McCloskey, B. Ahmed, J. Legg, and D. P. Hughes, "Deep Learning for Image-Based Cassava Disease Detection," Frontiers in Plant Science, Original Research vol. 8, no. 1852, 2017.
[25]S. Sladojevic, M. Arsenovic, A. Anderla, D. Culibrk, and D. Stefanovic, "Deep Neural Networks Based Recognition of Plant Diseases by Leaf Image Classification," Computational Intelligence and Neuroscience, vol. 2016, p. 3289801, 2016.
[26]M.-L. Zhang and Z.-H. Zhou, "A Review on Multi-Label Learning Algorithms," IEEE transactions on knowledge and data engineering, vol. 26, no. 8, pp. 1819-1837, 2013.
[27]Z. Min-Ling and Z. Zhi-Hua, "A K-Nearest Neighbor Based Algorithm for Multi-Label Classification," in 2005 IEEE International Conference on Granular Computing, vol. 2, pp. 718-721, 2005.
[28]M.-L. Zhang and Z.-H. Zhou, "Multilabel Neural Networks with Applications to Functional Genomics and Text Categorization," IEEE transactions on Knowledge and Data Engineering, vol. 18, no. 10, pp. 1338-1351, 2006.
[29]G. Tsoumakas, I. Katakis, and I. Vlahavas, "Random K-Labelsets for Multilabel Classification," IEEE Transactions on Knowledge and Data Engineering, vol. 23, no. 7, pp. 1079-1089, 2010.
[30]G. Tsoumakas and I. Vlahavas, "Random K-Labelsets: An Ensemble Method for Multilabel Classification," in European conference on machine learning, pp. 406-417: Springer, 2007.
[31]M. R. Boutell, J. Luo, X. Shen, and C. M. Brown, "Learning Multi-Label Scene Classification," Pattern recognition, vol. 37, no. 9, pp. 1757-1771, 2004.
[32]J. Read, B. Pfahringer, G. Holmes, and E. Frank, "Classifier Chains for Multi-Label Classification," in Joint European Conference on Machine Learning and Knowledge Discovery in Databases, 2009, pp. 254-269: Springer.
[33]J. Read, B. Pfahringer, G. Holmes, and E. Frank, "Classifier Chains for Multi-Label Classification," Machine learning, vol. 85, no. 3, p. 333, 2011.
[34]R. Caruana, "Multitask Learning," Machine learning, vol. 28, no. 1, pp. 41-75, 1997.
[35]Y. Sun, Y. Chen, X. Wang, and X. Tang, "Deep Learning Face Representation by Joint Identification-Verification," in Advances in neural information processing systems, pp. 1988-1996, 2014.
[36]A. Kendall, Y. Gal, and R. Cipolla, "Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics," in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 7482-7491, 2018.
[37]J. Dai, K. He, and J. Sun, "Instance-Aware Semantic Segmentation Via Multi-Task Network Cascades," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 3150-3158, 2016.
[38]K. Hashimoto, C. Xiong, Y. Tsuruoka, and R. Socher, "A Joint Many-Task Model: Growing a Neural Network for Multiple Nlp Tasks," in Conference on Empirical Methods in Natural Language Processing, 2017.
[39]M. Ji, K. Zhang, Q. Wu, and Z. J. S. C. Deng, "Multi-Label Learning for Crop Leaf Diseases Recognition and Severity Estimation Based on Convolutional Neural Networks," Soft Computing, vol. 24, no. 20, pp. 15327-15340, 2020.