Paddy plant leaf diseases identification using machine learning approach

Abstract

Agriculture is the most important sector in the Indian Economy and gives contribution in the form of agricultural productivity. To increase the agricultural productivity, precise and on-time detection of crop diseases and pest is needed. Most of the times, farmers fail to take the necessary steps even if they may have been able to identify the problem. Moreover, in some rural areas farmers cannot get rid of these problems because they do not have proper knowledge or education on how to do so. Most of the cases involve leaf diseases which are not recognized properly, and farmers end up using insecticides which may not be suitable for that particular disease. This paper provide new techniques of image pre- processing and a new combination of feature set from the processed images to make a trained model that will show how accurate our methodology is to detect diseases accurately. In this work, Machine learning algorithms namely KNN and SVM are implemented which can detect leaf diseases accurately. Among various plant leaf diseases, Rice leaf disease is one of them. This work is based on three Rice leaf diseases, they are - Bacterial leaf blight, Leaf smut, Brown spot.

Deep Convolutional Neural Networks Fusion with Support Vector Machines and K-Nearest Neighbors for Precise Crop Leaf Disease Classification

“…M. Micheni et al [28] carried out an experiment on maize data set using AlexNet and ResNet-50 with the help of transfer learning along with SVM, amounting accuracies of 98.3%, 96.6% and 88.5% respectively. Paddy leaves were used by Naware et al [29] to classify diseases using KNN and SVM giving 96.2% and 98.56% accuracies respectively. A. Nigam et al [30] proposed a new method for paddy leaf images classification using Principal Component Analysis (PCA) and Bacterial Foraging Optimization Algorithm (BFOA) with cost function for feature extraction and deep neural network used for classification to get an accuracy of 98%.…”