AUTOMATED PLANT DISEASE DETECTION SYSTEM FOR EARLY IDENTIFICATION OF TOMATO LEAF DISEASE

Abstract

This invention relates to an enhanced method for detecting diseases in tomato plant leaves using a deep learning model based on the DarkNet53 architecture. The system leverages transfer learning and advanced optimization techniques (SGDM, ADAM, and RMSProp) to achieve high classification accuracy across multiple tomato leaf disease categories. The framework ensures robust detection with minimal computational overhead and provides superior precision, recall, and F1-scores for ten distinct classes of diseases, including bacterial spot, early blight, late blight, and yellow leaf curl virus. The method demonstrates test accuracy up to 99.30% with AUC values nearing 1.0, establishing a reliable solution for early disease identification in agricultural practices.

Specification

Description:1. 1. Leaf Disease Dataset
The process begins with a dataset containing images of tomato leaves affected by various diseases.
2. Data Splitting
The dataset is divided into three subsets:
 Training Data
 Validation Data
 Test Data
3. Pre-processing
Pre-processing is applied to both training and test data to enhance the quality of the images and prepare them for model input. This may include:
 Image resizing
 Normalization
 Augmentation
4. Training Phase
 Enhanced Deep Learning (DL) Classifiers: Deep learning models (e.g., DarkNet53) are trained using the pre-processed training data.
 Training Options (Optimizers): Optimizers and other hyperparameters are configured to improve model performance during training.
 Validation Data: Used to fine-tune the model and avoid overfitting.
5. Class of the Disease
The trained model predicts the disease category for each input image during the training phase.
6. Testing Phase
 Pre-processed test data is used to evaluate the final model performance.
 The model classifies unseen test images and outputs the predicted disease class.
, Claims:1. A method for detecting tomato plant leaf diseases using an enhanced DarkNet53 deep learning model, comprising:
 Inputting images of tomato leaves into a pre-trained neural network.
 Employing transfer learning techniques to adapt the pre-trained network for the classification of tomato leaf diseases.
 Enhancing the model's performance through the use of advanced optimization algorithms, including SGDM, ADAM, and RMSProp.
2. The system of claim 1, wherein the model achieves a test accuracy of up to 99.30% across ten distinct disease categories.
3. The method of claim 1, wherein the network classifies tomato leaf diseases with precision and recall values exceeding 98% for the majority of disease categories.
4. The system is configured to generate real-time disease classification reports with an area under the curve (AUC) value approaching 1.0, ensuring high diagnostic reliability.
5. A non-transitory computer-readable medium containing instructions for executing the method described in claim 1, facilitating early detection and intervention in the management of tomato crops.
6. The system of claim 1, wherein specific diseases, including bacterial spot, early blight, and yellow leaf curl virus, are detected with 100% precision and recall when the ADAM optimizer is utilized.

Documents