Structural attributes are fundamental biophysical parameters of forests, useful for environmental monitoring and plan-ning. Canopy height (CH) is an important input for estimating several biophysical parameters such as aboveground biomass and carbon stocks, and can be associated with forest degradation, deforestation, emission reduction. Thus, an accurate CH estimation is a crucial issue in climate change, helping to increase biomass estimation accuracy, and support REDD+ initiatives. Very-high-resolution (VHR) imagery from unmanned aircraft systems (UAS’s) have been studied as a low cost means for CH estimation at local scales, however, estimation the accuracy is a factor that deter-mines its effectiveness. We evaluated the ability of VHR imagery from UAS’s to derive structural attributes, specifically tree-crown area and height, in a tropical forest fragment located in the foothills of the Andes, in the humid tropical forests of the region known as Biogeographic Chocó in Colombia South America. We used a structure from motion (SfM) approach to derive the forest fragment’s CH, and we applied mean-shift algorithms to identify single tree crowns. We performed accuracy assessment using tree height derived from field campaigns and visual interpretation of VHR imagery. Results showed a RMSE of 3.6 m of the canopy height model (CHM) with a R2 = 0.75; the total accuracy for delineating tree crowns was 73.9%. We found that using VHR imagery collected by UASs, specific trees and canopy gaps can be identified in forest fragments, which is an important step to determine forest structure.