Downloads: 1
Research Paper | Computer Science | Volume 15 Issue 7, July 2026 | Pages: 947 - 956 | India
A Comprehensive Analysis of Attention Mechanisms for Fine-Grained Coconut Leaf Disease Detection
Abstract: Leaf diseases of plants constitute an alarming threat to crop productivity, especially with reference to coconut cultivation. Bulk inspection of such plant leaves is extensively done by visual examination, which is extremely time-consuming. State-of-the-art advancements in the field of deep learning have introduced plant disease inspection using computer vision, while most existing techniques are based on general plant leaves, i.e., they are not plant-specific, thus failing to provide the required information. This study investigates the impact of attention mechanisms on fine-grained coconut leaf disease detection using an EfficientNet-based architecture. In this work, two attention mechanisms, namely Squeeze-and-Excitation (SE) and Convolutional Block Attention Module (CBAM), are used to emphasize channel and spatial feature representation in an EfficientNet baseline model. Our models are trained and tested on a coconut leaf disease dataset under realistic conditions. An extensive ablation study is performed by comparing the baseline model against the attention-enhanced variants. Experimental results show that attention-based models reach a better classification performance, with the CBAM-enhanced EfficientNet outperforming others on accuracy, precision, recall, and F1-score. Moreover, Grad-CAM visualizations are used to highlight disease-relevant regions, enabling model interpretability. Based on the obtained results, it may be inferred that attention mechanisms significantly enhanced feature learning and improved the classification performance in the task of fine-grained coconut disease detection. This makes the proposed approach suitable for practical agricultural applications.
Keywords: Coconut leaf disease detection, EfficientNet, Attention mechanisms, SE block, CBAM, Fine-grained classification, Grad-CAM, Deep learning in agriculture
How to Cite?: Anil Kumar R J, Sumanashree Y S, Dr Siddaraju K, Nirmala M S, "A Comprehensive Analysis of Attention Mechanisms for Fine-Grained Coconut Leaf Disease Detection", Volume 15 Issue 7, July 2026, International Journal of Science and Research (IJSR), Pages: 947-956, https://www.ijsr.net/getabstract.php?paperid=SR26709150835, DOI: https://dx.doi.org/10.21275/SR26709150835