Hybrid CNN & Random Forest Model for Effective Clove Leaf Disease Dignosis

Clove (Syzygium aromaticum) farming is at the crossroads of history and innovation, producing a versatile spice prized for its culinary or medicinal uses. This study investigates the use of sophisticated methods of machine learning for automated categorization of illnesses affecting clove leaves. Our model, which uses neural network convolution (CNN) or random forest methods, has remarkable precision, recall, or F1-Score for various disease classifications. Clove Rust, Clove Leaf Spot, Clove Anthrax, Clove Powdery Mildew, Clove Leaves Blight, Clove Bacterial Leaf Spot, Clove Fusarium Wilt, or Clove Septoria Leaf Spot are all rigorously analyzed in the categorization model. Precision ratings range from 88.12% to 92.23%, indicating that the model is accurate in recognizing cases of each condition. The model’s ability to minimize false negatives and capture all relevant instances is shown in its high recall values, which range from 87.38% to 90.83%. The F1-Score varies from 88.58% to 91.41%, indicating the model’s overall efficacy and a harmonic equilibrium between precision and recall. The number of cases for each illness class is represented by support metrics, which reveals how it varies across the dataset. The model obtains an amazing general precision of 97%, demonstrating its ability to correctly classify occurrences across a wide range of conditions. The model’s solid performance is further consolidated by macro, weighted, or micro averages, emphasizing its flexibility to varied datasets. The weighted average, which takes into account class imbalances, displays values of 89.73%, confirming the algorithm’s accuracy in diverse dataset configurations. The micro-average strongly matches with individual class measures, offering an in-depth evaluation of the model’s general efficacy. Finally, the combination of CNN with Random Forest is useful in increasing automated illness detection in clove leaves.