Explainable Deep Neural Network Based Multiclass Classification Of Tomato Leaf Disease

• Author(s):

  Rupesh Kumar | Dr. Ranu Pandey

• Keywords:

  Tomato Leaf Disease Detection, Hybrid Deep Learning, Lightweight Convolutional Neural Network, Lightweight Vision Transformer, Precision Agriculture, Explainable Artificial Intelligence, Grad-CAM, Image Classification.

• Abstract:

  Tomato Leaf Diseases Pose A Persistent Challenge To Sustainable Crop Production, Requiring Accurate And Computationally Efficient Diagnostic Solutions For Real-time Field Applications. This Study Proposes A Hybrid Deep Learning Framework That Combines A Lightweight Convolutional Neural Network (CNN) With Lightweight Transformer Architecture For Robust Classification Of Tomato Leaf Diseases. The CNN Component Is Employed To Extract Fine-grained Local Spatial Features, While The Transformer Module Captures Global Contextual Dependencies, Enabling Improved Discrimination Between Visually Similar Disease Categories. The Model Was Trained And Evaluated On A Multi-class Tomato Leaf Dataset Comprising Healthy And Diseased Samples, Including Bacterial Spot, Early Blight, Late Blight, Leaf Mold, And Septoria Leaf Spot. Data Augmentation And Transfer Learning Strategies Were Applied To Enhance Generalization And Mitigate Overfitting. The Proposed Hybrid Model Achieved An Overall Classification Accuracy Of 99.08%, With A Precision Of 98.93%, Recall Of 98.95%, And F1-score Of 98.94%. Comparative Analysis Indicates Superior Performance Over Standalone Lightweight CNN And Transformer Models, While Maintaining Reduced Computational Complexity Suitable For Resource-constrained Environments. To Improve Model Interpretability, Gradient-weighted Class Activation Mapping (Grad-CAM) Was Utilized To Visualize Disease-relevant Regions, Confirming That The Model Focuses On Meaningful Pathological Features. The Results Demonstrate That The Proposed Approach Provides A Reliable, Efficient, And Interpretable Solution For Automated Plant Disease Detection, Supporting Its Applicability In Precision Agriculture And Smart Farming Systems.

---

Other Details

• Paper id:

  IJSARTV12I5105415

• Published in:

  Volume: 12 Issue: 5 May 2026

• Publication Date:

  2026-05-21

Download Article