Plant Leaf Disease Detection using Deep Learning

Abstract

Pests damage plants and crops, which has an impact on the nation\'s agricultural output. Typically, farmers or professionals use their own eyes to monitor the plants to look for disease and identify it. However, this approach could be time-consuming, expensive, and unreliable. Results from automatic detection employing image processing methods are quick and precise. In this study, deep convolutional networks are used to develop a novel method of classifying leaf images in order to recognise plant diseases. The technique of precise plant protection has the potential to grow and improve, and computer vision advancements have the potential to boost the market for applications in precision agriculture. Innovative training methods and the methodology employed make it simple and quick to implement the system in real-world settings. The deep convolutional neural network used in this method paper has been trained and fine-tuned to fit accurately to a database of plant leaves that was gathered independently for various plant illnesses. The innovation and advancement of the proposed model lay in its simplicity; by utilising deep CNN, the model can discriminate between ill and healthy leaves as well as between them and the environment. Healthy leaves and backdrop images are also in line with other classes.

Introduction

I. INTRODUCTION

Agriculture is one of India's key economic sectors. The Indian agricultural sector employs about 60% of the labour force of the nation. The largest producer of pulses, rice, wheat, spices, and spice-related items is believed to be India. The quality of the items that farmers produce, which is mostly dependent on the plant, determines how successful their businesses are. Plants are quite susceptible to illnesses that stunt their growth, which in turn has an impact on the farmer's environment. Use of automatic disease detection techniques is beneficial for spotting plant diseases at their earliest stages. In some sections of a plant, such as the leaves, the symptoms of plant diseases are obvious. It is laborious to manually diagnose plant illness using photographs of the leaves. Therefore, it is necessary to create computer techniques that would automate the disease identification and categorization procedure using leaf images.

Viral, fungal, and bacterial illnesses including Alternaria, Anthracnose, bacterial spot, canker, etc. are the principal diseases that affect plants. The bacterial disease is caused by the presence of germs in leaves or plants, the viral disease is caused by environmental changes, and the fungus disease is caused by the presence of fungus in the leaf. The process of segmentation is based on various aspects of an image, such as colour orientation, texture, borders, etc. Image segmentation is the process of dividing a picture into various parts. In this study, a Gradient Boosting Algorithm is used to segment leaves. When image processing is used for automatic illness identification, less work is required, costs are low, and on the plus side, it takes less time and is more accurate. In order to detect and identify plant diseases at an early stage and improve product quality, this study examines the significance of image processing techniques.

A. Motivation

The main motivation for producing this project is that, while every section of the world is developing, there is no such huge achievement or development in plant leaf diseases. So if we can prioritise this leaf field and detect infections, it will be useful to them.

B. Problem Statement

Veggies are very vulnerable to plant-impairing illnesses development that affects farmers' livelihoods ecology. Utilizing an automated illness detection system method is useful for finding plants early stages of illness Plant ailments appear themselves throughout the plant, including the leaves. The process of manually diagnosing takes a while utilising images of leaves to diagnose plant illness. The result is Development of computational algorithms is necessary to automate the disease identification procedure, and classifying with the aid of  leaf photos.

II. RELATED WORKS

Dhruvi gosai; binal kaka; dweepna garg; radhika patel et.al. The ResNets algorithm was the focus of this paper. A component of the artificial neural network is a residual neural network (ResNet) (ANN). The vanishing/exploding gradient issue can be resolved using the residual block in the ResNet technique. Residual Networks are also created using the ResNet technique. ResNets produce a substantially better outcome for picture classification. The ResNets techniques used several of the parameters, including weight decay, gradient clipping, and scheduling learning rate. The researchers anticipate highly accurate outcomes and identifying more diseases from the varied harvests using the ResNet algorithm.

Divyanshu Varshney; Burhanuddin Babukhanwala et.al.This article's emphasis was on leaf plant disease. Using a transfer learning approach like deep learning, a new method for detecting plant leaf diseases has been created. CNN is utilised as a feature extractor and SVM is used for classification. The suggested model's evaluation was conducted using the benchmark dataset PlantVillage. The recommended model performed better than earlier studies, obtaining an 88.77 percent training accuracy, when it was analysed and compared to current approaches.

Shima Ramesh; Ramachandra Hebbar; Niveditha M. et.al. The phases of implementation included in the proposed article are dataset construction, feature extraction, classifier training, and classification. To categorise the photos of sick and healthy leaves, the produced datasets of sick and healthy leaves are combined and trained using Random Forest. Utilizing the Histogram of an Oriented Gradient, we may extract characteristics from a picture (HOG). Overall, we can clearly identify the illness present in plants on a massive scale by utilising machine learning to train the vast data sets that are publically available.

Amrita S. Tulshan; Nataasha Raulet. Et.al. You can use this method to identify a disease from the supplied photographs of plant leaves. Steps including picture pre-processing, image segmentation, and feature extraction were part of this procedure. On the results of these three phases, the K Nearest Neighbor (KNN) classification is applied further. The proposed implementation's accuracy in forecasting plant leaf diseases is 98.56%. Additionally, it includes the Affected Area, Disease Name, Total Accuracy, Sensitivity, and Elapsed Time for a plant leaf disease.

III. FLOW DIAGRAM

Provide a plant leaf picture input format as part of the input dataset to the machine. Data Preprocessing: The dataset is first processed manually using Python modules. Preparing the raw data to be acceptable for a machine learning model is known as data pre-processing. The following stage is feature extraction, which is a dimensionality reduction procedure that divides a large initial collection of raw data into smaller, more manageable groups for processing. Classification - In supervised Machine Learning (ML), classification is the act of determining the class or category of data based on previously established data classes that have been "labelled." Utilize ML to classify the model. After completing the procedures, the output is the detection of plant leaf disease.

IV. ALGORITHM

Convolutional Neural Network

Convolutional neural networks with a focus on image and video recognition applications. Cnn is primarily utilised for image analysis applications such segmentation, object detection, and picture recognition.

Convolutional neural networks have four different kinds of layers:

1. Convolutional Layer: Each input neuron in a conventional neural network is connected to the following hidden layer. Only a small portion of the input layer neurons in cnn are connected to the hidden layer of neurons.
2. Pooling Layer: The pooling layer is used to make the feature map less dimensional. Inside the cnn's hidden layer, there will be numerous activation and pooling layers.
3. Flatten: Flattening is the process of reducing data to a 1-dimensional array so that it may be entered into the following layer. We flatten the convolutional layer output to produce a solitary, lengthy feature vector.
4. Fully Connected Layer: Fully connected tiers make up the network's final few layers. The output from the last pooling or convolutional layer is passed into the fully connected layer, where it is flattened before being applied.

Conclusion

With the help of a combination of shape, texture, and colour feature withdrawal, this idea intended to highlight disease in the leaf. Farmers first provide a digital photograph of a plant\'s damaged leaf, which is read into the system by image preprocessing and automatically analysed using CNN algorithm. The goal of this study is to produce data that are useful for identifying sick leaves caused by specific plant diseases that are frequently seen.

References

[1] Dhiman Mondal, Dipak Kumar Kole, Aruna Chakraborty, D. Dutta Majumder’ Detection and Classification Technique of Yellow Vein Mosaic Virus Disease in Okra Leaf Imagesusing Leaf Vein Extraction and Naive Bayesian Classifier., 2015, International Conference on Soft Computing Techniques and Implementations- (ICSCTI) Department of ECE, FET, MRIU, Faridabad, India, Oct 8-10, 2015. [2] Pranjali B. Padol, Prof. AnjilA.Yadav, ‘SVM Classifier Based Grape Leaf Disease Detection’ 2016 Conference on Advances in Signal Processing(CAPS) Cummins college of Engineering for Women, Pune. June 9-11, 2016. [3] Aakanksha Rastogi, Ritika Arora, Shanu Sharma, “Leaf Disease Detection and Grading using Computer Vision Technology &Fuzzy Logic,” presented at the 2nd International Conference on Signal Processing and Integrated Networks (SPIN), IEEE, 2015, pp. 500–505. [4] Garima Tripathi, Jagruti Save, “AN IMAGE PROCESSING AND NEURAL NETWORK BASED APPROACH FOR DETECTION AND CLASSIFICATION OF PLANT LEAF DISEASES,” Int. J. Comput. Eng. Technol. IJCET, vol. 6, no. 4, pp. 14–20, Apr. 2015. [5] S. Arivazhagan, R. Newlin Shebiah, S. Ananthi, S. Vishnu Varthini, “Detection of unhealthy region of plant leaves and classification of plant leaf diseases using texture features,” Agric Eng Int CIGR J., vol. 15, no. 1, pp. 211–217, Mar. 2013. [6] Ms.Poojapawer ,Dr.varshaTukar, prof.parvinpatil ‘Cucumber Disease detection using artificial neural network’ [7] Detecting jute plant disease using image processing and machine learning 2016 3rd International Conference on Electrical Engineering and Information Communication Technology (ICEEICT) [8] An Individual Grape Leaf Disease Identification Using Leaf Skeletons and KNN Classification (2017) N.KRITHIKA, DR.A.GRACE SELVARANI. [9] Hierarchical Learning of Tree Classifiers for Large-Scale Plant Species Identification.(2015) Jianping Fan, Ning Zhou, Jinye Peng, Ling Gao. [10] Plant Disease Detection Using Leaf Pattern:A Review(2015) Vishnu S, A. Ranjith Ram. [11] Detection of Diseases on Cotton Leaves Using K Mean Clustering Method(2015) Pawan P. Warne, Dr.S.R. Ganorkar.

Copyright

Copyright © 2023 S. V. Shinde, Hrithik D. Dhyade, Yash Meshram, Swapnil Mergel, Nirmal Neve. 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.