SVM based Leaf Disease Classification Assisted with Smart Agrobot for the Application of Fertilizer

Abstract

Agriculture is playing prominent role in day to day life of every individual. Food is the basic need of every living organism in the world. Agriculture builds and gives life to biotic environment. For quality agriculture it is essential to ensure that soil, plants and leaves are disease free. The proposed work is based on the background of improving the quality of agriculture. This work describes an approach to detect diseases in few of the leaves namely Hibiscus, Neem, Curry Leaf which we grow more often in small fields and houses nearby. The proposed system uses Machine learning model in conjunction with Agrobot (Agricultural Robot) where the detection of disease is followed by appropriate fertilizer spray. The proposed system works on the basis of IoT technology in order to facilitate the communication between agrobot and the farmer. Machine learning model uses SVM (Support Vector Machine) to classify the disease followed by action taken by agrobot towards application of fertilizer spray. Here the application of IoT technology in the hardware setup of proposed system plays indeed an important role in monitoring moisture, temperature and humidity level in sensor through appropriate sensors and it is frequently communicated with farmer. The camera module in the hardware setup captures the images of leaves and the testing phase of SVM (Support Vector Machine) is executed to get the result as either healthy leaf or detected disease. The application of agrobot is to ensure the spray of necessary fertilizer at required time so as to prevent the occurrence and spread of leaf diseases. The calculation of accuracy is done for the obtained results and observed that accuracy in classification increases with increased number of samples in SVM Training.