Detection of Plant Disease using Deep Learning

With the increase in world population over the years, the demand for food has increased exponentially. Moreover, Agriculture with its allied sectors is the largest source of livelihood in India. Plant diseases are a major risk factor for food security. Hence it becomes very important to detect the plant diseases in crops accurately and quickly to reduce the damage caused to the crops. Generally, the identification of plant diseases and remedies for those disease is done by experts. This requires the farmers to bear the transportation cost for reaching the experts and the consultation fees charged by the experts for disease detection. To reduce these problems caused to the farmers, they can use automatic plant disease identification. In this process, one can take a photograph of the diseased plant leaves and upload them to our servers. On servers, we use a trained deep learning model (Convolutional Neural Network architecture) to detect the disease in the plant leaf. Then the detected disease is displayed to the user along with the remedy for the disease. The Neural network was trained using more than twenty thousand images of diseased plant leaves and was tested on more than three thousand six hundred images of diseased plant leaves. The trained neural network was found to be able to predict the disease in the leaves with a good accuracy of 97%. This will help reduce all the above-mentioned costs incurred by the farmers and will also save the time and energy of the farmers. In addition to this automatic detecting is more accurate than visual inspection by the experts. It will also keep the historical data and analysis accessible at any time thus improving the handling of the challenges faced due to plant diseases. In this way, our project is aimed towards reducing the hardships faced by the farmers and providing accurate results.