Jepri Banjarnahor, Reclesia Br Harianja, Syafridatul maulidah, Nenda Sartika Manalu

Hybrid Deep Learning Model for Coffee Leaf Disease Detection Using CNN DeiT

Introduction

Hybrid deep learning model for coffee leaf disease detection using cnn deit. Detect coffee leaf diseases accurately using a hybrid deep learning model combining CNN and DeiT. Achieves 93.1% accuracy, robust, and scalable for precision agriculture.

Abstract

Coffee production plays a crucial role in the agricultural economy; however, its productivity is significantly affected by plant diseases that are difficult to detect at early stages. Accurate disease identification remains challenging due to subtle visual differences and high intra-class variability in leaf symptoms. To address this problem, this study proposes a hybrid deep learning framework that integrates Convolutional Neural Networks (CNN) and Data-efficient Image Transformers (DeiT) for automated coffee leaf disease classification. The proposed architecture leverages CNN to capture fine-grained local features, while DeiT models global contextual relationships through self-attention mechanisms, enabling a more comprehensive feature representation. The model is trained and evaluated on a dataset of 6,048 labeled images across four classes: Healthy, Rust, Red Spider, and Leaf Miner. Experimental results demonstrate that the proposed CNN–DeiT model outperforms baseline CNN and Transformer-based approaches, achieving an accuracy of 93.1%, an F1-score of 92.3%, and a ROC-AUC of 95.6%. Robustness analysis shows that performance degradation remains limited (1.6%–3.4%) under various perturbation conditions, while out-of-distribution evaluation indicates strong generalization capability with only a minor accuracy decrease. These findings confirm that the hybrid CNN–Transformer architecture effectively enhances classification performance, robustness, and generalization. This study contributes to the advancement of deep learning methodologies in agricultural image analysis by providing a robust and scalable framework for plant disease classification, with potential applications in precision agriculture and data-driven crop management.

Review

This study tackles a crucial problem in agricultural economics: the early and accurate detection of coffee leaf diseases, which significantly impact productivity and yield. The authors effectively highlight the inherent challenges in this domain, specifically the subtle visual differences between disease symptoms and high intra-class variability, underscoring the pressing need for advanced automated solutions. This clear problem statement and motivation set a strong foundation for the proposed research into an innovative deep learning framework designed to overcome these limitations. To address these challenges, the paper introduces a novel hybrid deep learning model that integrates Convolutional Neural Networks (CNNs) with Data-efficient Image Transformers (DeiT). This architectural choice is well-justified, leveraging the strengths of CNNs in extracting fine-grained local features and DeiT's proficiency in capturing global contextual relationships via self-attention mechanisms, thereby enabling a more comprehensive feature representation. The model's performance is rigorously evaluated on a substantial dataset of 6,048 labeled images across four critical classes. Impressively, the proposed CNN-DeiT model demonstrates superior results, achieving a 93.1% accuracy, 92.3% F1-score, and 95.6% ROC-AUC, consistently outperforming standalone CNN and Transformer-based approaches. The inclusion of robustness analysis, showing minimal performance degradation under perturbations, and strong generalization capabilities in out-of-distribution scenarios further bolsters the credibility and practical applicability of the proposed method. The findings confirm that this hybrid CNN–Transformer architecture significantly enhances classification performance, robustness, and generalization in the context of plant disease detection. This study provides a valuable contribution to the field of deep learning methodologies in agricultural image analysis by presenting a robust and scalable framework with clear potential for real-world application. Its implications for precision agriculture and data-driven crop management are substantial, offering a promising avenue for improving coffee production efficiency and sustainability. The work is well-executed, addressing a significant problem with an innovative and thoroughly evaluated solution.

Full Text

Comments

You need to be logged in to post a comment.