ARTIFICIAL INTELLIGENCE IN CROP DISEASE DETECTION

Abstract

The world’s population is increasing day by day today we stood at the more than 7.9 billion population mark while in case of India is the world second most populated country of the world with the population of more than 1.4 billion and these numbers are increasing day by day, food is the very basic need for the survival and the diseases are the most common threat to the crop like cereal crop, cash crop, fruits, vegetables that hampers the yield from the crop, early prediction of the disease is required so that farmers can get the full yield amount from the crop. Remote sensing with incorporating Artificial Intelligence gives a solution of this problem by early detecting the disease, in this research the crop mapping is performed with the help of Supervised Classification like Maximum Likelihood Classifier and Random Forest classifier in open-source software Quantum GIS with the help of Semi-Automatic Classification plugin, the overall accuracy achieved in case of Maximum Likelihood Classification is 68.39 % with the kappa coefficient of 0.5868 and in case of Random Forest the overall accuracy is received is 79.7437 % with the kappa coefficient of 0.7129 making the Random Forest the most suitable algorithm for the crop mapping in the heterogenous agriculture farm land. There are 3 major classes identified in the field as Wheat, Sugarcane, and other crops like Tree (mango orchards), Popular, Mustard and Vegetables, the wheat crop produces the producer’s accuracy of 79.20% and user accuracy of 84.00% in the random forest classification method which is substantial in case of Sentinel-2 satellite data, after identifying the wheat crop in the field disease detection are performed on the temporal data of sentinel – 2 in the complete cropping period by the help of vegetation indices. Soil Adjusted Vegetation Index and Normalized Difference Vegetation Index are calculated for the study area, and it is found that the SAVI is very sensitive than NDVI by producing clear separation in values of index specially in the NIR zone of wavelength, SAVI is further analyzed for the disease mapping and a trend find for the marked healthy and diseased fields find out in the temporal data.