Plant Disease Prediction using Deep Learning and IoT

Crop diseases are a major threat to food security, but due to lack of infrastructure this area remains untouched. We are proposing a system based on IoT and Deep Learning. This system would try to identify the stress to the plants by soil fertility, environmental imbalance and crop diseases. The system would also try devising a plan to overcome it. There are already some existing systems to solve the similar problem. But the major shortfall of these systems is that they only consider the leaf part and try to detect the crop disease. Our proposed system would consider the leaves but also check the environmental stress and soil fertility by measuring moisture, salinity, etc. The crops would monitored by drones fixed with cameras and the images would be captured at regular interval of time. The system would work on datasets of Bottle Gourd, Maize and Papaya with healthy and diseased leaves which are captured in controlled environment. From all these captured data, machine would be trained based on Convolutional Neural Network. Then all the captured data would be tested on the trained machine. The system would predict whether the leaves are affected by disease or not. Further any changes in weather or sudden change in environment would also be tracked through Web API's. All these results would be bundled and reports would be generated. Overall reason behind moving towards neural networks was due to large datasets and accurate results. It is being observed that neural networks provide good result on massive datasets.