Enlightening Plant Leaf Identification through 2D CNNs with Dual Attention Mechanism

Plant leaf identification plays a pivotal role in various fields, including agriculture, botany, and environmental monitoring. Leveraging advancements in deep learning, this study proposes a novel approach for plant leaf identification using 2D Convolutional Neural Networks (CNNs) augmented with a Dual Attention Mechanism. The proposed model aims to enhance the discriminative capabilities of traditional CNNs by selectively focusing on informative regions within leaf images. The Dual Attention Mechanism operates at both spatial and channel levels, allowing the model to dynamically attend to relevant features while suppressing irrelevant ones. Through this mechanism, the model learns to capture intricate patterns and subtle variations present in diverse leaf structures, thereby improving classification accuracy and robustness. We evaluate the performance of our approach on publicly available leaf datasets and compare it with existing methods. Experimental results demonstrate that the proposed model achieves superior accuracy and outperforms state-of-the-art techniques in plant leaf identification tasks. Furthermore, we conduct extensive ablation studies to analyze the contribution of the Dual Attention Mechanism and validate its effectiveness in enhancing the model's discriminative power. Overall, this study highlights the potential of integrating attention mechanisms with 2D CNNs for advancing plant leaf identification techniques, paving the way for more efficient and accurate automated systems in botanical research and environmental conservation efforts.