Analysis on Prediction of Plant Leaf diseases using Deep Learning
S. Nandhini, K. Ashokkumar
Abstract
As the global population continues to burgeon, increasing overall crop production is becoming imperative to ensure food safety for everyone. However, a myriad of plant diseases can sever the supply of essential crops. To tackle these diseases, it is first important to identify these diseases and monitor them on a large scale. To solve this problem, Tensorflow with Keras and OpenCV are used to build a detection system. Since automated farms tend to be large and spread out, the classifier has to be executed and controlled from a cloud computing environment. This system can process images of plants and detect common diseases. Using three convolution layers, ten nodes and forty epochs this model was able to achieve validation of more accuracy. The training set consisted of 11942 images and the validation set was thirty-five per cent of the training set in size i.e a total of 6421 images. It is believed that the detection systems like ours, can strengthen the farming industry and help secure food safety for all. OpenCV allows our system to detect patters in images and translate those patterns into data on which ML models can be built using Keras. While previous attempts at solving similar problems have been made, few models that can specifically target Indian plants have been made, especially with a dataset of this size and a model of forty epochs.