Investigation of Deep Learning Methods for Disease Detection in Cotton Fields using UAV Imagery

Cotton, an important crop for growing textile sector, has been in decline with constant trends. This research primarily focuses on developing an improved system for detecting and classifying diseases in cotton fields using deep learning techniques applied to Unmanned Aerial Vehicle (UAV) imagery. As UAV deals with nonlinear behavior in flight control and detects disease in cotton field using UAV Imagery constantly poses serious challenges. The challenges include, Image noise, camera motion problem, pictorial distortions and issue of lighting conditions at various locations. The research aims at accurately detecting cotton diseases using deep learning through the help of UAV Imagery. The UAV is aimed at flying over the field to gather the data and make sure that the image gather has better resolution to make use for disease detection in larger sense. The images include healthy, diseased and severely diseased class. The 7 fields data at different times and location are gathered and are subjected to preprocessing. This preprocessing aims at achieving good accuracy and functions in the best possible manner. Many algorithms of deep learning are tested to evaluate their accuracy and precision in terms of disease detection. This project is scalable and can be extended to other crops, enhancing agricultural productivity.