Plant Disease Classification and Identification Using Deep Convolutional Neural Network

Abstract

The first step in effectively and accurately preventing plant disease in a complicated environment is to identify the diseased plants. Given the rapidity development is occurring. The recognition of diseased plants is made computerised and data-driven as a result of smart farming, enabling enhanced assistance with decisions, clever analysis, and management. Plant Disease has developed widely, necessitating precise investigation and quick deep learning applications in the classification of plant diseases. When a plant is sick, it may exhibit a variety of signs and symptoms, such as coloured streaks or spots on the leaves. The colour, structure, and proportions of the visual symptoms alter as the disease worsens. Deep CNN can be used to train these patterns and produce a model that can identify them. The framework was designed and developed using the MobileNet V2 architecture. A 94% accuracy rate was reached after training on a chosen dataset. Using the chosen topology for 30 epochs results in the development of a convolutional neural network. For this undertaking, a user interface may also be designed. Farmers can use any smartphone with an adequate camera to take pictures of plant leaves by utilising the application. Multiple disorders can be identified using the smartphone application, making it simpler for users. Along with the diagnosis of plant diseases, pesticide recommendations and precautions for the recognised diseases are also recommended. This research offers a deep learning method for automating the detection of plant diseases. Convolutional neural networks (CNN) are used in the created system.