Plant Leaf Disease Detection Using CNN Algorithm

Literature Review

The foremost step in the process of software development is the literature survey. Determining the economic strength and time factor is essential before creating any software for the problem statement in hand. Plants are the primary source of energy (Cseke et al.,2009) and have a prominent role in solving Global Warming. Diseases are threatening the livelihood of this significant food source (Strange et al.,2005). Object recognition and image classification problems are much solved by different types of algorithms.

(Serawork et al.,2018) identified soya bean plant diseases using Convolutional Neural Networks (CNN). The authors have studied the feasibility of CNN for plant leaf disease classification using the images taken from the environment. They used LeNet architecture to create a model that classifies the soya bean plant diseases by using a dataset consisting of 12,673 images of healthy and diseased leaves belonging to four classes that were acquired from the PlantVillage database. The Dataset includes images taken under different conditions of the environment. The model implemented proves that CNN can obtain features of the leaves and classify diseases with an accuracy of 99.32% from the images of plants in various conditions. The authors also mentioned that there is an imbalance in the data set classification and suggested implementation of batch normalization for speeding up the process and improving accuracy.