Census of individual trees of Hacienda San Jose Vichada (2023)

This dataset contains information of estimated individual trees in Hacienda San José (HSJ), Vichada-Colombia extracted from remote sensing imagery processed using AI models and the TreeEyed QGIS Plugin. The dataset includes a CSV file with attributes related to tree dimensions and geolocation and a shapefile with polygons representing each tree.This dataset contains information of estimated individual trees in Hacienda San José (HSJ), Vichada-Colombia extracted from remote sensing imagery processed using AI models and the TreeEyed QGIS Plugin. The dataset includes a CSV file with attributes related to tree dimensions and geolocation and a shapefile with polygons representing each tree.

Metodology:The tree information for this dataset was obtained using a custom QGIS plugin (TreeEyed) and python scripting, designed to process high-resolution RGB remote sensing imagery to derive tree data. Leveraging AI pretrained models, in this case Meta’s HighResCanopyHeight model was employed. The methodology involved processing RGB imagery with a resolution of 0.5 meters and dimensions of 42,686×30,630 pixels for the region of interest, a subdivision in 1260 tiles of 1024×1024 pixels to process and obtain canopy tree height rasters, merging, vectorizing tree instances, and filtering to isolate individual trees. Zonal statistics were then applied to assign estimated tree height values to each instance by extracting the maximum pixel value within the tree crown polygon boundary. This approach allowed for flexible and localized analysis, enabling the derivation of detailed tree metrics such as height, crown area and perimeter, and equivalent diameter from high-resolution RGB imagery.

Ruiz Hurtado, A.F.; Pérez Bolaños, J.; Arrechea-Castillo, D.A.; Cardoso Arango, J.A