"UAV chirpine Murree"

"This study mainly focuses on counting the Chir Pine trees by using UAV imagery. Methods of traditional individual tree detection and modern deep learning models were applied to find out which model is best for forest monitoring. High resolution UAV images were used for analysis while traditional ITD methods were created from three-dimensional point clouds which are generated by using Digital Terrain Models (DTM) and Digital surfaces Models. (DSM). Deep learning methods such as YOLO, Transformer Yolo and deep forest are applied directly to the Ortho mosaic images. Results show that ITD methods produced moderate accuracy by a precision of 0.64%, recall of 0.73% with F1 score of 0.68%. The custom trained deep forest and transformer custom trained YOLO achieved high accuracy. While transformer custom trained deep forest shows the greatest performance amid total accuracy of 93.33%. This result shows that deep learning methods are more accurate, especially in complex and dense forest areas. Deep learning models can directly use UAV images which save time and provide reliable results and best for Chir pine forests."