A Unified Approach to Tomato Leaf Disease Recognition: CNN and Random Forest Integration

The analysis presented here summarizes a thorough assessment of a classification model's efficacy in identifying several plant diseases, highlighting accuracy, recall, and F1-s across multiple classes. Notably, each disease class has high precision values that vary from 91.67% to 94.70%, including Early Crime, Fusarium Wilt, The powdery Mildew, Anthracnose and Septoria Leaves Spot, Botrytis, Microbial Speck, and Southern Blight. These scores represent the model's precision in recognizing positive occurrences for each illness category, with Bacterio Speck obtaining the highest precision. The recall values, which represent the model's capacity to catch genuine cases, consistently vary between 90.45% to 93.75%, highlighting the model's efficacy across a wide spectrum of disorders. The F1-scores, which include precision and memory, show consistent excellence, fluctuating between 91.32% to 93.77%, demonstrating a balanced achievement in obtaining high recall as well as accuracy.The Macro Average, calculated as the free-of-weight mean of memory, accuracy, and the F1-s throughout all classes, yields 92.85%, 92.84%, as well as 92.84%, accordingly. This demonstrates the model's consistent efficacy across the classification spectrum. The Weighted Average enhances the model's trustworthiness by delivering identically weighted averages while accounting for class dispersion. The Micro Average, which takes into account the pooled counts of genuine positives, erroneous positives, and erroneous negatives, yields an average precision of 92.85%. Furthermore, the reported overall accuracy rating of 92.8422% provides a summed appraisal of the model's capacity to accurately categorize instances spanning all diseases, giving stakeholders a comprehensive picture of how well it is performing.