IoT and AI Based Integrated System to Detect Crop Diseases and Deficiency of Nutrients in a Large Farm

Abstract

The world population is expected to increase by 2 billion by 2050. A person dies of hunger every 3.6 seconds. The UN has come up with a list of 17 goals to make the world a better place and ending hunger comes at second. A 119% increase in yield is required by 2050 to sustain life. As the amount of land cannot be increased, the only way is by increasing crop yield from the same land. Research has shown that disease and nutrient deficiency-related yield losses have cost the world 60% of global agricultural productivity.  Diagnosis of the disease and nutrient deficiency in a large farm is very difficult because of the number of crops. This project, “CropMates”, aims to identify crop diseases in a vast area of farmland by first surveying the acreage using a GPS-enabled drone. The camera-fitted drone captures video and pictures of the farm while flying over the field. A deep learning algorithm is created to identify leaves from these pictures. Another deep learning-based algorithm using AlexNet architecture is trained with 38 different classes of diseases on 87,000 images of leaves to identify diseases from these leaves.  An Artificial Neural Network is developed and is trained on 20,000 data points to identify nutrient deficiencies in the soil from IoT-enabled sensors put across the farm. An app is built to show the results and to recommend the type and amount of pesticide and fertilizer for optimum crop yield.