Drone Photos Map Trees in Malunda Natural Forest: Accuracy Assessed

Image: article cover, from Irlan et al. (2025), Jurnal Wasian, licensed CC BY 4.0.

A research team from the University of Sulawesi Barat has tested individual tree mapping in the natural forest of the Malunda Forest Management Unit (KPH) in West Sulawesi, using aerial photographs from RGB camera drones. This method is considered far cheaper than using LiDAR sensors, which are commonly employed for forest mapping. The results show that the approach is feasible, although it still tends to overcount the number of trees.

In the study published in Jurnal Wasian, the team recorded 614 sample trees in 30 sample plots of 20 x 20 metres and compared them with the mapping results from drone photos. A point cloud-based algorithm successfully separated 683 trees. Its accuracy reached a recall of 0.68, precision of 0.76, and F-score of 0.72. The positions of detected trees deviated by an average of 1.95 metres from their actual positions, while the crown radius deviated by 1.59 metres. The overcounting occurred because dense canopy cover made field measurements incomplete, so small trees and undergrowth were also detected as tree crowns.

For forest managers, these findings offer a more affordable way to inventory trees in natural forests with complex structures. Manual inventory has so far required significant time, cost, and labour. The research team states that improving the quality of RGB point clouds and tree separation algorithms will be the focus of future research.

Source and how to cite

This story is based on the research article Individual Tree Segmentation in TropicalNatural Forest Based on Point CloudGenerated from UAV RGB Image by Irlan Irlan, Ulfa Adzkia, Suhartono Suhartono, Meliani Meliani, Alpri Sri Jenos, Teguh Bimantara and Chairil A, published in Jurnal Wasian Vol. 12 No. 02 (2025). The full article is free to read on the Jurnal Wasian website.

Cite the original article (APA 7):

Irlan, I., Adzkia, U., Suhartono, S., Meliani, M., Jenos, A. S., Bimantara, T., & A, C. (2025). Individual Tree Segmentation in TropicalNatural Forest Based on Point CloudGenerated from UAV RGB Image. Jurnal Wasian, 12(02), 78–93. https://doi.org/10.62142/kx7bhn83

If you use these findings in coursework, a thesis or other writing, cite the original article above, not this news story. The original article is licensed CC BY 4.0.