Fine-Tuned CNN-Based Approach for Multi-Class Mango Leaf Disease Detection

2025-10-06

AI Summary

To address the low accuracy and poor generalizability in multi-class mango leaf disease identification in South Asian agriculture, this study proposes an intelligent diagnostic method based on transfer learning and model fine-tuning. We systematically evaluate five pre-trained CNNs—DenseNet201, InceptionV3, ResNet152V2, SeResNet152, and Xception—on eight mango leaf diseases. DenseNet201 demonstrates superior robustness and discriminative capability under complex agricultural conditions. After fine-tuning, it achieves a test accuracy of 99.33% with balanced F1-scores across all classes, notably outperforming competing models in distinguishing morphologically similar diseases—namely, cutworm damage and bacterial canker. This work validates the efficacy of densely connected architectures for fine-grained plant disease classification and establishes a reproducible, cost-effective technical pathway for high-reliability field-deployable intelligent diagnosis.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Mango is an important fruit crop in South Asia, but its cultivation is frequently hampered by leaf diseases that greatly impact yield and quality. This research examines the performance of five pre-trained convolutional neural networks, DenseNet201, InceptionV3, ResNet152V2, SeResNet152, and Xception, for multi-class identification of mango leaf diseases across eight classes using a transfer learning strategy with fine-tuning. The models were assessed through standard evaluation metrics, such as accuracy, precision, recall, F1-score, and confusion matrices. Among the architectures tested, DenseNet201 delivered the best results, achieving 99.33% accuracy with consistently strong metrics for individual classes, particularly excelling in identifying Cutting Weevil and Bacterial Canker. Moreover, ResNet152V2 and SeResNet152 provided strong outcomes, whereas InceptionV3 and Xception exhibited lower performance in visually similar categories like Sooty Mould and Powdery Mildew. The training and validation plots demonstrated stable convergence for the highest-performing models. The capability of fine-tuned transfer learning models, for precise and dependable multi-class mango leaf disease detection in intelligent agricultural applications.

Problem

• Develops fine-tuned CNN models for mango leaf disease classification
• Compares five pre-trained architectures for multi-class disease identification
• Evaluates model performance using accuracy and other standard metrics

Innovation

• Fine-tuned CNN models for mango disease detection
• DenseNet201 achieved highest accuracy of 99.33%
• Transfer learning strategy with five pre-trained networks