Plant disease identification using Deep Learning: A review

Abstract

The paper reviews various classification techniques exclusively used for plant disease identification. Early stage plant disease identification is extremely important as that can adversely affect both quality and quantity of crops in agriculture. For identification of plant diseases, different approaches like image processing, machine learning, artificial neural networks, and deep learning are in use. This review focusses on an in-depth analysis on recently emerging deep learning-based methods starting from machine learning techniques. The paper highlights the crop diseases they focus on, the models employed, sources of data used and overall performance according to the performance metrics employed by each paper for plant disease identification. Review findings indicate that Deep Learning provides the highest accuracy, outperforming existing commonly used disease identification techniques and the main factors that affect the performance of deep learning-based tools. This paper is an attempt to document all such approaches for increasing performance accuracy and minimizing response time in the identification of plant diseases. The authors also present the attempts for disease diagnosis in Indian conditions using real dataset.