dScout : Unmanned Ground Vehicle for Automatic Disease Detection and Pesticide Atomizer

Abstract

Agriculture Cyber-Physical System (A-CPS) is intensively gaining significance in achieving precision agriculture. Plant diseases are the major problem in the agricultural sector, which affects crop production and, in return, affects the economic profit of India as most people are dependent on agriculture. In conventional agriculture, manual monitoring of crops is laborious, slow, and error-prone process to be implemented. The economic and production loss due to plant diseases can be tackled with continuous plant monitoring. Contributing towards precision agriculture, the proposed system (dScout) is able to automate the real-time problem of leaf disease detection with deep computer vision and the Internet of Things. This paper deals with the automation of image acquisition using Raspberry pi and smart line tracking robot, identification of leaves with YoloV3 CNN architecture, the spread of infection with image processing, and disease detection with InceptionV3, and automatic pesticide sprayer. The trained InceptionV3 model has achieved 90.51% accuracy and can differentiate nine tomato leaf diseases. The automatic pesticide sprayer is integrated with Raspberry pi, which predicts the diseases in images with the SVM model. The results are synchronized to the cloud and mobile interface. dScout provides a high-performance working solution for automatic crop disease detection and detailed disease prevention suggestions under the real agricultural environment.