An Efficient Crop Field Disease Prediction Using A Novel Parallel Mini Patch Aggregation Neural Network (Pmpann)

Abstract

The major sources for gross domestic product among the developing countries across the world is agriculture. The nature and amount of crop cultivated along with its yield in each region varies with soil and climate. Additionally, the crop yield is affected by the disease. Paddy is the major food crops cultivated in several parts and it succumbs to several diseases. The farmers often find it hard to predict the nature of disease over the paddy crop. For supporting the farmer, a novel Deep Convolution Mini Patch Aggregation Neural Network (DCMPANN) classifier is proposed to predict and localize the disease in paddy crop using its image. The images of crop are collected in real-time and the novel hybrid tanh and sigmoid function is developed for the proposed classifier. Using performance metrics the performance of the DCMPANN model has been evaluated. The proposed DCMPANN model demonstrates the effectiveness on disease prediction on validated with existing systems.