Intelligent plant disease diagnosis using convolutional neural network: a review

In recent times use of different technologies for intelligent crop production is growing. To increase the production of crops, diagnosing a plant disease is very important. Plant diseases can be identified using various techniques like image processing, machine learning, deep learning, etc. Among these techniques deep learning, especially deep learning using convolutional neural networks (CNN) has proved to be more efficient in recent years compared to other methods. This manuscript focuses mainly on the diseases affecting on eleven (11) different plants and how the diseases can be identified from plant leaf images using CNN based deep learning models. This review can help the researchers to get a brief overview of how state-of-the-art CNN models can be used for disease diagnosis in plants, and an overview of the state-of-the-art studies that have used visualization techniques to identify the disease spots for better diagnosis. The review also summarises the studies that have used hyperspectral images for plant disease diagnosis and various data sources used by different studies. The challenges that currently exist while developing a plant disease diagnostic system and the shortcomings and open areas for research have also been discussed in this manuscript.

Automated summary

The use of various technologies for intelligent crop production is increasing, with deep learning using convolutional neural networks (CNN) proving to be more efficient for plant disease diagnosis. This review focuses on the identification of plant diseases from leaf images using CNN based deep learning models and provides an overview of state-of-the-art studies using visualization techniques and hyperspectral images for disease diagnosis. The challenges and areas for further research in developing a plant disease diagnostic system are also discussed.