International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 05 | May 2022
p-ISSN: 2395-0072
www.irjet.net
Plant Disease Detection using Convolution Neural Network (CNN) Nita patil1, Alpesh Tandel2, Rushika Gawade3, Arati kamble4 1Assistant
professor, Computer Engineering Datta Meghe College of Engineering, Airoli, New Mumbai – 400708(Maharashtra) (India) 2,3,4Students, Computer Engineering Datta Meghe College of Engineering, Airoli, New Mumbai – 400708(Maharashtra) (India) ---------------------------------------------------------------------***--------------------------------------------------------------------aspects/objects in the image and be able to differentiate one Abstract— When crop plant is suffering from pests it attacks the agricultural production of the world. As usual farmers and experts focus the plants by eye for detect and notice of disease. But this manual process is time processing, high-cost and inexact. Therefore, there is need for accurate and efficient automatic detection of the plant diseases. The main aim of this project is to find a solution to the problem of 38 different classes of plant diseases detection using the simplest approach while making use of minimal computing resources to achieve better results and high accuracy compared to the traditional models. Convolution Neural Network (CNN) training model is deployed for detection of plant diseases. CNN model employs automatic feature extraction to help in the classification of the input image into respective disease classes. This proposed system has achieved an average accuracy of 92% indicating the feasibility of the neural network approach even under unfavorable conditions. Key Words: CNN, Disease detection, Confusion Matrix
1. INTRODUCTION The agriculture production of the farmer is much reduced if crops are having pests. Manual process of identification and recognition of diseses by experts is cumbersome and takes lot of time. Previous technologies based on feature extraction from the images of the plant have less accuracy as compared to deep learning approaches. l All steps are required for implementing this disease recognition model are fully expressed in this project, starts from collecting images and information to make a database, evaluate by agricultural experts, a deep learning substructure to achieve the deep CNN learn. This project may be a new perspective in detecting plant diseases using the deep convolutional neural network trained and refine to become accurately to the database of a plant’s leaves that was converged unaccompanied for diverse plant diseases. The proposed model is detecting the disease from the given plant leaf image. This can be achieved using CNN. In the proposed model we have used Convolution Neural Network (CNN) for the classification of plant disease detection. Convolutional Neural Network (CNN) is a Deep Learning algorithm which can take in an input image, assign importance (learnable weights and biases) to various © 2022, IRJET
|
Impact Factor value: 7.529
|
from the other. Using CNN for plant disease detection is not a new domain. A lot of work has been already done on this domain. But we have found out that most of the research used very small dataset as well as the activation functions that they used can be replaced with large dataset and various optimization functions. So, the proposed model will be having improved the accuracy of plant disease detection using CNN over the previous work by applying newer technique. We have also created a simple website giving the image as an input and the website will classify the Plant disease detection using the proposed CNN model. The dataset that we have used for this project consists of plant leaf images which were taken from Kaggle dataset.
2. LITERATURE SURVEY Prasanna Mohanty et.al, 2016 [1], detected disease in plants by training a convolutional neural network. CNN model is trained to classify healthy and unhealthy plant of 14 crops. This model achieved an accuracy of 99.35% on test dataset. Malvika Ranjan et.al, 2017 [2], proposed use HSV features for feature extraction on cotton plant and used Artificial Neural Network (ANN) to classify disease crops and healthy samples. The ANN model achieved an accuracy of 80%. S. Arivazhagan et.al, 2013 [3], proposed model process involves four main process as follows first, a color transformation structure is take as input RGB picture, and then it means of a specific threshold value then green pixels are detected , which is followed by segmentation steps, and for obtaining advantages of segments the texture statistics are calculated. At the end, classifiers are used for features that are taken to identify the disease. Kulkarni et.al, 2017 [4], applied image processing methods to identify the plant diseases for right and accurate detection of plant diseases using artificial neural network (ANN) and diverse image processing methods. As the proposed approach is based on ANN classifier and Gabor filter for feature extraction, it got better output with a rate of recognition is 91%. R.P Narmadha et.al, 2017 [5], applied Image processing techniques which can be defined as the technical analysis of an image by detecting the disease of plants leaf using complex algorithms of machine learning. B R , Jagdesh et.al, 2019 [6], proposed methodology in which histogram matches and their histogram specification is the transformation image so that their histogram matches the ISO 9001:2008 Certified Journal
|
Page 1309