Efficient Coastal Mangrove Species Recognition Using Multi-Scale Features Enhanced by Multi-Head Attention

• DOI: 10.3390/sym17030461
• Publication Date: 2025-03-19T11:48:37Z
• Authors: Shaolin Guo; Yixuan Wang; Yin Tan; Tonglai Liu; Qin Qin

Recognizing mangrove species is a challenging task in coastal wetland ecological monitoring due to the complex environment, high similarity, and inherent symmetry within structural features of species. Many coexist, exhibiting only subtle differences leaf shape color, which increases risk misclassification. Additionally, mangroves grow intertidal environments with varying light conditions surface reflections, further complicating feature extraction. Small are particularly hard distinguish dense vegetation their symmetrical that difficult differentiate at pixel level. While hyperspectral imaging offers some advantages recognition, its equipment costs data acquisition complexity limit practical application. To address these challenges, we propose MHAGFNet, segmentation-based recognition network. The network utilizes easily accessible RGB remote sensing images captured by drones, ensuring efficient collection. MHAGFNet integrates Multi-Scale Feature Fusion Module (MSFFM) Multi-Head Attention Guide (MHAGM), enhance improving capture across scales integrating both global local details. In this study, also introduce MSIDBG, dataset created using high-resolution UAV from Shankou Mangrove National Nature Reserve Beihai, China. Extensive experiments demonstrate significantly improves accuracy robustness recognition.