Abstract

India’s economy is vastly based on agriculture. Increase in the productivity of agricultural fields affects the GDP of the nation. Early and automated crop disease detection is an important step to increase the productivity of these farms. This paper proposes a model using machine learning classifiers for the detection and classification plant disease and is based on tomato plant but similar results can also be obtained using other crops. Experimental results also reveal that 76 % and 72% of accuracy level obtained using Extremely Randomized Tree classifiers and Random forest outperformed other machine learning classifiers. Though our model not produced higher accuracy when compared to other pre-trained Convolutional Neural Network (CNN) models such as Res net and Image Net, the difference in time complexity made our model less memory extensive.

Key words: Agriculture, Disease Identification, Machine Learning, Image Processing, Random Forest, KNearest Neighbours.