Potato Leaf Disease Detection using Sequencial Models

Potatoes are the third most important food source in the world after wheat and rice. Moreover, India, in the fiscal year 2021, produced over 50 million metric tons of potatoes out of which _ percent was wasted due to late blight and early blight diseases. The traditional methods to detect diseases in the plants involve manual inspection. This method is very expensive, time-consuming and does not provide satisfactory results. So as to identify the infected leaves at the beginning of their growth cycle which will help to increase the yield and thereby decrease the losses incurred by the farmers, we propose a web application to do the same with the help of deep learning models like InceptionNet, ResNet50, MobileNet and CNN trained on the PlantVillage dataset available on kaggle. We achieved an accuracy of 93.97 percent, 88.79 percent, 96.12 percent and 94.83 percent respectively.