Plant Disease Detection and Crop Recommendation Using CNN and Machine Learning

The wide-scale prevalence of diseases in crops and inefficient soil to grow crops highly damage the standard quality and quantity of crop production. So, the disease in the crops needs to be early diagnosed by developing or employing a fast and innovatory approach and crop recommendation system will benefit the farmers. Hence, this study proposed a system that has the ability to detect diseases in plants using CNN as well as recommend various crops based on the quality of the soil by performing analysis on its various parameters using ML. The dataset for disease prediction training and test is obtained from the Plant Village Dataset and correctly separated and therefore various species of plants are recognized and re-named to make an accurate database. The next step is to obtain a test database that will be consisting of different diseases in plants that are used to check the accuracy and confidence level of the proposed module. Then the classifier is trained using training data and after that, the output is going to be detected with the best accuracy. And for the crop recommendation system, the Support vector classifier (SVC) algorithm is used as it outperforms compared to other classifiers like KNN, Logistic Regression, Random Forest, and Decision Trees, in the system to improve the efficiency rate of our model. The developed model also maps the soil and crop database and suggests suitable crops based on the available nutrients level of the soil and thus allows formers to make better decisions regarding the type of crops that can be sown-in in the field. This study also compared the performance of various classifiers on the available dataset for study and chose the one with the highest accuracy.