Urban forest management requires accurate tree detection to maintain ecological balance in densely populated areas. This study presents an approach to enhance tree accuracy by integrating the YOLOv8 machine learning model with a region-growing algorithm, leveraging high-resolution RGB UAV data. The methodology uses a Canopy Height Model (CHM) to precisely detect individual trees, particularly in dense urban areas with complex canopy structures. The current approach is validated on data from Maharashtra, achieving a tree detection accuracy of 75% in dense regions. These results highlight the potential of integrating machine learning models with image processing techniques to achieve accurate tree detection in challenging urban settings. This approach enables urban planners and managers to optimize resource allocation, monitor forest health, and develop data-driven strategies for sustainable urban forest management.