Machine Learning for Apple Orchard Health: CNNSVM Model for Disease Detection

A disease identification model's performance in apple farm disease management is evaluated in this study. The study targets five primary apple leaf diseases: Scab, Fire Blight, Cedar Rust, Powdery Mildew, and Leaf Spot. This study assesses the model's disease classification accuracy and reliability using advanced performance indicators. Each illness class has high precision values, 89.40% to 95.24 %. The model's precision scores show its capacity to limit false positives, proving its disease prediction accuracy. The model has outstanding recall rates of 90.91% to 93.65%, indicating its capacity to recognize all illness cases. The model's reliability and efficacy are reinforced by the F1-scores' balanced evaluation (91.47%–93.02%). Overall model accuracy of 92.19% is shown by strong cumulative findings across all illness classifications. The model's disease predictions are accurate across 6508 photos, proving its reliability. The model's micro-average performance of 92.19% shows its ability to generalize and identify diseases in varied situations. The model's consistent performance across all illness classes is shown by macro-averaged precision (92.47%), recall (92.09%), & F1-score (92.26%). The model's precision (92.26%), recall (92.19%), & F1-score (92.20%) weighted average values demonstrate its accuracy when class distribution changes.