Hybrid CNN & Random Forest Model for Effective Curry Leaf Disease Diagnosis

This abstract evaluates the efficacy of a model of classification in identifying and classifying different diseases of plants. To measure the model's efficacy across multiple illness classes, recall, precision, F1-score, support, percentage, or accuracy metrics were used. Notably, the model identified "Leaf Spot Diseases" (84.03%) or "Powdery Mildew (Erysiphe spp.)" (77.98%) with excellent precision and recall rates of "Leaf Curl Diseases" (74.45%) or "Viruses" (69.09%). The weighted average measures (precision: 85.40%, retention: 86.00%, F1-score: 84.90%) demonstrate the system's general robustness in dealing with uneven class distributions. The macro average numbers (precision: 70.85%, remembering: 67.45%, and F1-score: 70.11%) show consistent and equitable results across all classes. This study emphasises the model's efficacy for early illness identification and provides insight into effective agriculture disease management tactics.