Individual Tree Detection for Qualitative Inventory Eucalyptus Pellita Using Unmanned Aerial System (UAS)

The increasing application of remote sensing technologies, particularly the use of Unmanned Aerial Systems (UAS), has significantly advanced forest resource assessment. Advances in remote sensing data acquisition technology, particularly Unmanned Aerial System (UAS), have expanded the knowledge of remote sensing in forestry to produce high-resolution images and affordable 3D data. This study aims to extract and automatically calculate the accuracy of individual tree detection for industrial plantation forests ( Eucalyptus pellita ) using the local maxima algorithm for Canopy Height Model (CHM) data and to assess silvicultural quality based on tree height. The study was conducted in industrial plantation forest areas located in Tapung District, Kampar Regency, Riau. Seven field plots measuring 4 x 4 trees were manually counted in the field. A total of 99 trees were successfully detected out of 112 trees in the entire plot, resulting in an accuracy of 88% (F-Score 0.89). Overall, the algorithm used missed 17 trees, incorrectly detected 5 trees, and correctly detected 93 trees, resulting in recall and precision values of 0.95 and 0.84, respectively. The COV, Gini, CRR, and PV50 indices yielded values of 0.71, 0.096, 0.12, and 7.57, respectively. Overall, the tree height distribution was relatively uniform with low variation and a moderate concentration of values.