Prediction of grape leaf through digital image using FRCNN

Abstract

A great deal of research indicates that the duality of agricultural production can be reduced because of various factors. Plant diseases are among the most critical aspects of this category. As a result, reducing plant diseases allows for significant improvements in quality. The article uses Grape Leaf Disease Detection Technique (GLDDT) with Faster Region based Convolutional Neural Network (FRCNN) - GLDDT-FRCNN techniques to automatically diagnose plant diseases. Once trained, the software can diagnose plant leaf disease without requiring any experimentation. The primary focus of this research is grape leaf disease. The basic technique manipulates H & colour histograms 25 channels 24. Excluding the final 26 phases, where the person decides which channel (H or a) offers the best separation, the algorithm method is fully automated. A GLDDT is also proposed in this article, which uses two-pronged processes for the evaluation, recognition and categorization of traits. The analysis process, testing on a benchmark set of data reveals that the disease diagnosis system might be a better fit than existing methods because it recognizes and identifies infected/diseased areas. The researchers achieved a precision rate of 99.93% for the detection of Isariopsis, black rot and Esca using the proposed disease detection method.

Keywords

• Grape leaf disease
• CNN algorithm
• Grape leaf disease detection technique
• Classification