Evaluation of Tree Diameter and Height Measurements in UAV Data by Integrating Remote Sensing and Machine Learning Methods

This study evaluates the effects of different ground sampling distances on the diameter and height measurements of brutian pine trees in point cloud data from unmanned aerial vehicle photographs. The study is located within the Çandır Forest Management Directorate of the Isparta Regional Directorate of Forestry. The results serve as independent variables in machine learning methods to predict field-measured diameter and height values. Nine distinct machine learning techniques were used, including AdaBoost Regression, Artificial Neural Networks, Deep Neural Networks, Decision Tree Regression, Gradient Boosting Regression, Linear Regression, Random Forest Regression, Support Vector Regression, and eXtreme Gradient Boosting Regression. The results show that predictions made using data with a low ground sampling distance had the lowest correlation values for diameter and height, while predictions made using data with a high ground sampling distance had the lowest correlation values. Deep Neural Network achieved the highest success rate for diameter estimation, while Decision Tree Regression had the lowest success.