Plant Disease Detection: A Comparative Study of Deep Learning Approaches

Demand for food is increasing every year. It causes insecurity in food items. Farmers are using large amounts of fertilizers and pesticides to meet their needs. Fertilizers are substances applied to plants to increase their productivity. According to the survey, India ranks second in fertilizer use globally and first in south Asian countries. Pesticides are used to kill and repel certain kinds of pests. The overuse of pesticides and fertilizer causes a lot of problems. Most pesticides are harmful and have a terrible effect on the environment and human health. Sulphur, Malathion, Mutchler, Copper Oxychloride, Prorate, and Mancozebs have commonly used pesticides. Overuse of pesticides causes severe health issues like Cancer and that leads to death. Managing and controlling pesticides is very important to ensure safe public health. There are numerous reasons for the over usage of pesticides. The Main reason is crop diseases and pest attacks. Inefficient ways to determine plant disease and late diagnosis cause more use of pesticides. Determining the correct plant disease and types is essential to overcome this threat. This work deals with an effective way to discover plant disease detection using deep learning models. Numerous learning methods are present for the early detection and classification of diseases. In this work, we apply different deep learning algorithms such as CNN, ResNet, Inception model, Vgg16 and Vgg19. We get an accuracy in the range of 72% to 92%.