Simulated Web Application for Peanut Leaf Diseases Detection And Classification

Abstract

This project presents a leaf diseases detection and classification for the peanut plant using machine learning. This system can be used to identify early and late diseases spots on the peanut plant leaf. The arranged higher psychological cycle framework uses picture content portrayal and regulated classifier style of neural organization. Picture measure procedures for such a call examination includes preprocessing, highlight extraction and an arrangement stage. At the cycle, partner input pictures will be resized utilizing an open CV and the area of interest picked. Here, shading related surface alternatives square measure removed from the contribution for network training and characterization. Shading choices like mean, the difference of HSV Color territory and surface alternatives like energy, differentiation, homogeneity and relationship. The framework will be acquainted with arrange the check pictures consequently to choose leaf qualities. For this methodology, programmed classifier CNN is utilized for arrangement upheld learning with some instructing tests of that class. This organization utilizes digression sigmoid to work as piece work. The proposed system will be a research tool for the study of early and late spots occurring on the peanut plant and it will be trained using an online platform called Kaggle and the proposed system can be used by the researchers via a web portal (web site). The final web portal is used for generating dynamic graphs and for publishing research findings in the feed.