Date of Publication :1st June 2023
Abstract: Diseases affecting plants contribute to declining crop yields and associated economic costs. The cost, duration, and precision of a plant disease diagnosis depend on the early detection methods. This study analyzes the many techniques for identifying plant diseases by examining available photos and using different processing algorithms. It does this by utilizing both traditional machine learning methods and deep learning methods to perform a careful analysis of the work that has been done in the literature about the datasets that were used, the various image processing techniques that were implemented, the models that were used, and the efficiency that was obtained. The paper explains each technique's potential pitfalls and advantages and the obstacles that need to be overcome for efficient plant disease diagnosis. The results suggest that deep learning is superior to other machine learning algorithms in identifying plant diseases, whereas visible-range photos are preferred over spectral ones
Reference :