Tomato leaves diseases detection approach based on support vector machines
Download as PPTX, PDF•2 likes•4,580 views
The document presents a research approach to detect diseases in tomato leaves using support vector machines and gabor wavelet transform techniques. It outlines the challenges faced in manually identifying plant diseases, emphasizing the need for automated detection to reduce economic losses and improve agricultural quality. Experimental results showed high classification accuracy (up to 100%) for different kernel functions, indicating the effectiveness of the proposed method.
Tomato leaves diseases detection approach based on support vector machines
- 1. Tomato leaves diseases detection approach based on support vector machines The 11th International Computer Engineering Conference (ICENCO2015) – Cairo, Egypt Usama Mokhtar http://www.egyptscience.net 1
- 2. Agenda Introduction Problem Definition Motivation Proposed Approach Experimental Results Conclusion and Future Works 2 The 11th International Computer Engineering Conference (2015)
- 3. Tomatoes are one of the most widely cultivated food crops throughout the world due to its high nutritive value. It contains a lot of vitamins and nutrients such that vitamin C. It occupies the fourth level between word vegetables. Egypt is one of the famous countries that interested in tomatoes cultivation. It ranked fifth among leader countries in the world. Introduction 3 The 11th International Computer Engineering Conference (2015)
- 4. During cultivation process, tomato leaves expose to many of problems and diseases such as: Late blight powdery mildew Early blight Bacterial spot Gray mold Problem Definition 4 The 11th International Computer Engineering Conference (2015)
- 5. Problem Definition The naked eye observation of experts is the main approach adopted in practice for detection and identification of plant diseases. But this approach requires continuous monitoring of experts which might be expensive and difficult especially in large farms. 5 The 11th International Computer Engineering Conference (ICENCO2015) – Cairo,
- 6. Problem Definition So, It is necessary to help farmers in automatically detect symptoms of disease as soon as they appear by analysing the digital images which may helping us for: minimizing major production and economic losses, ensuring both quality and quantity of agricultural products and minimizing agrochemicals use. 6 The 11th International Computer Engineering Conference (2015)
- 7. Motivation The aims of this research are: To present a hybrid model that employs gabor wavelet transform technique to extract relevant features related to image of tomato leaf along with Support Vector Machines (SVMs) with alternate kernel functions in order to detect and identify type of disease that infects tomato plant. 7 The 11th International Computer Engineering Conference (2015)
- 8. General proposed approach The 11th International Computer Engineering Conference (2015) 8
- 9. Image preprocessing phase 9 The 11th International Computer Engineering Conference (2015)
- 10. Feature extraction phase 10 The 11th International Computer Engineering Conference (2015)
- 11. Classification phase 11 The 11th International Computer Engineering Conference (2015)
- 12. Overall structure layout 12 The 11th International Computer Engineering Conference (2015)
- 13. Experimental Results: Used Dataset The datasets used for experiments were constructed based on real sample images of diseased tomato leaves. Datasets of total 200 infected tomato leaf images with Powdery mildew and early blight were used for both training and testing phase. In this approach, SVM is employed using different kernel functions including Cauchy kernel, Invmult Kernel and Laplacian Kernel. Grid search and N-fold cross-validation techniques were used to parameters selection and performance evaluation of the 13 The 11th International Computer Engineering Conference (2015)
- 14. Experimental Results 14 The 11th International Computer Engineering Conference (2015)
- 15. Experimental Results cont… 15 The 11th International Computer Engineering Conference (2015) test result for invmult_kernel
- 16. Experimental Results cont… 16 The 11th International Computer Engineering Conference (2015) test result for Laplacian_kernel
- 17. Experimental Results cont… 17 The 11th International Computer Engineering Conference (2015) test result for Cauchy_kernel
- 18. Conclusion and Future Works Experimental results indicated that the proposed approach outperformed the typical SVMs classification algorithm with different classification accuracies as follow: 78% for Invmult kernel functions. 98% for Laplacian kernel function, accuracy is increased by ≈ 20%. 100% for Cauchy kernel function, accuracy is increased by only 2%. 18 The 11th International Computer Engineering Conference (2015)
- 19. Conclusion and Future Works For future research, variety of challenges and research directions could be considered. Some general research directions are to consider more plant diseases with different conditions Another open problem is to tackle the second problem, which faces SVMs or any classification system; namely feature selection, using PSO. Moreover, a hybrid approach for optimizing SVMs parameters and select best features subset is planned to be developed. 19 The 11th International Computer Engineering Conference (2015)
- 20. References Peralta, E. Iris , and M. David Spooner. “History, origin and early cultivation of tomato (Solanaceae).” Genetic improvement of Solanaceous crops, vol. 2, pp. 1-27, 2006. Kong, Wai Kin, David Zhang, and Wenxin Li. “Palmprint feature extraction using 2-D Gabor filters.” Pattern recognition, vol. 36, no. 10, pp. 2339-2347, 2003. Chen, Hui-Ling, Bo Yang, Gang Wang, Su-Jing Wang, Jie Liu, and DaYou Liu. “Support vector machine based diagnostic system for breast cancer using swarm intelligence.” Journal of medical systems, vol. 36, no. 4, pp. 2505-2519, 2012. 20 The 11th International Computer Engineering Conference (2015)
- 21. Thanks and Acknowledgement 21 The 11th International Computer Engineering Conference (2015)