CNN and Logistic Regression for Accurate Mango Gummosis Severity Prediction

In this study, a model of Convolutional Neural Network (CNN) is proposed for classifying the ten Intensity Scales from 1-10% to91-% severity in leaf diseases. In the end note of a paper, they claimed that the model achieved remarkable accuracy at 99% in differentiating among stages of Gummosis disease. 12,405 photos made up the dataset, and the model's total accuracy of 92.62% was rather good. The precision rate of the model ranges at Intensity Scale 6 from as low as 96.15% up to Intensity Scale 9 with a high target accuracy detection for regressing specific levels of different disease intensities. Recall values lie between 96.27% to 98.31%, with higher performance in Intensity Scales {8,9}. F1-scores: These balance precision and recall, with scores ranging from 96.21% to 98.28%, peaking in the higher scales Number of instances per Intensity Scale (support) and Accuracy which ranges from 631 to 1832 supporting values at each scale, the accuracy is 100% for all scales. The excellent model results demonstrate the model capacity- providing accurate detection and classification of mango leaf disease which brings new heights in managing diseases in Mango crops.