Smart Agriculture: Leveraging CNN-RF Fusion for Enhanced Tomato Disease Classification

This paper presents a unique method to the categorization of tomato illnesses, using a combination of models involving Random Forest (RF) and Convolutional Neural Networks (CNN). The hybrid model incorporates both types of neural networks. The trial results show that the predictive algorithm has an extraordinary capacity to reduce the number of false positives, with accuracy values ranging from 78.7 to 95.1%. The model's strong recall scores, ranging between 86.1% to 95.2%, reflect its ability to recognize occurrences of each unique disease class and gather real positives. The model can detect true positives, confirmed by these studies. Furthermore, the balanced equilibrium of the F1 Ratings, which may range from 84.4% to 95.1%, demonstrates that the framework is capable of executing sickness classification tasks on a consistent and dependable basis. This new set of findings adds to the body of evidence. Statistics on sickness incidence are also accessible at various support levels, with some choices providing more precise data than others. The CNN-RF hybrid model is an effective and adaptable tool for tomato disease classification. This is due to the fact that it is able to reach accuracy values that are consistently between 0.96 and 0.98.