Forecasting Anthracnose Severity Levels in Mango Leaf using Hybrid Models

The aim of this study is to investigate the use of Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) to properly diagnose the degree of anthracnose illness in mango leaf samples. The models demonstrated remarkable performance metrics across a range of severity classifications, indicating their potential for reliable disease evaluation. Precision values show how well the models can reduce false positives and accurately identify each severity class, with values over 90% across all severity levels. Furthermore, strong recall rates (above 90%) in all severity levels highlight the models' ability to identify genuine positives among real positives, which is essential for precise disease diagnosis. The models' consistent performance in forecasting Anthracnose severity levels was confirmed by the F1-Score, which balances accuracy and recall and has consistently been around 92%. Furthermore, the models continually showed outstanding accuracy of 97%, indicating their trustworthiness in properly diagnosing the severity of anthracnose. The models' resilience was notably confirmed by weighted average, macro, and micro measures, which showed steady performance even in the presence of possible class imbalances in the dataset. These results highlight the potential of CNN and SVM models as useful instruments for accurate and timely evaluation of Anthracnose severity in agricultural settings, supporting researchers and farmers in putting targeted disease control plans into practice. Overall, the models' excellent accuracy and low misclassification rate highlight how important they are to improving agricultural disease assessment techniques and enhancing crop health and productivity.