<![CDATA[Accurate mapping of coastal ecosystems is essential for marine resource management and biodiversity conservation. This study integrates UAV (Unmanned Aerial Vehicle) technology, Agisoft Metashape photogrammetry, and Object-Based Image Analysis (OBIA) using eCognition to map mangrove ecosystems and benthic habitats in the coastal area of Tiwoho Village, North Minahasa Regency. Data acquisition using DJI drone produced 949 aerial photos at 60.1 meters altitude with 80% overlap covering an area of 0.361 km². Orthophoto with spatial resolution of 1.43 cm/pixel was generated through photogrammetric processing in Agisoft Metashape. Image classification was performed using a two-level hierarchical approach: Level 1 for general zonation (land, shallow water, deep water) using rule-based classification, and Level 2 for detailed benthic habitat identification (primary mangrove, secondary mangrove, sand, seagrass, coral) using Support Vector Machine (SVM) algorithm. Classification results showed Overall Accuracy of 89% and Kappa Index of 94.03% with User's Accuracy ranging from 0.89-0.97 and Producer's Accuracy 0.92-0.99. This integrated workflow proved effective for coastal ecosystem mapping with high accuracy and can be replicated in other regions. Keywords: UAV, photogrammetry, OBIA, Support Vector Machine, benthic habitat, mangrove]]>
Copyrights © 2026
