ORIGINAL RESEARCH article
Front. Plant Sci.
Sec. Sustainable and Intelligent Phytoprotection
Explainable Deep Learning-Based Comparative Study for Guava Fruit and Leaf Disease Classification: Advancing Agricultural Diagnostics through AI
Provisionally accepted
- Shandong Normal University, Jinan, China
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
Early detection of plant diseases is critical for maintaining crop health and ensuring agricultural productivity. In this study, we propose an explainable deep learning-based framework for the classification of guava fruit and leaf diseases. A real-world dataset comprising 527 annotated images across five disease classes—Disease Free, Phytophthora, Red Rust, Scab, and Styler and Root Rot—was utilized. We developed and compared six hybrid model architectures by combining transfer learning backbones (VGG16, MobileNetV2, InceptionV3, ResNet50) with custom CNN classifiers. Among these, the proposed VGG16 + MobileNetV2 hybrid achieved the highest performance with an accuracy of 96%, F1-score of 0.96, and strong class-wise generalization across all metrics. To ensure model transparency and enhance trust in AI-assisted agricultural diagnostics, we employed Gradient-weighted Class Activation Mapping (Grad-CAM) to visualize class-discriminative regions in diseased guava samples. Detailed visualizations including confusion matrices, ROC-AUC curves, PR curves, and radar plots validated the comparative strengths of each model. Our findings demonstrate the efficacy of combining deep and lightweight architectures for robust and interpretable disease classification. This approach holds significant promise for integration into real-time smart farming systems and mobile diagnostic tools, particularly in resource-limited agricultural environments.
Keywords: deep learning, Explainable AI, Grad-CAM, Guava disease classification, Hybrid CNN, Plant Pathology, smart agriculture, Transfer Learning
Received: 04 Nov 2025; Accepted: 13 Jan 2026.
Copyright: © 2026 Zhou. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Zeyu Zhou
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.