Unmanned aerial vehicles (UAV) are used successfully in many application areas such as military, security, monitoring, emergency aid, tourism, agriculture, and forestry. This study aims to automatically count trees in designated areas on the Siirt University campus from high-resolution images obtained by UAV. Images obtained at 30 meters height with 20% overlap were stitched offline at the ground station using Adobe Photoshop's photo merge tool. The resulting image was denoised and smoothed by applying the 3x3 median and mean filter, respectively. After generating the orthophoto map of the aerial images captured by the UAV in certain regions, the bounding boxes of different objects on these maps were labeled in the modalities of HSV (Hue Saturation Value), RGB (Red Green Blue) and Gray. Training, validation, and test datasets were generated and then have been evaluated for classification success rates related to tree detection using various machine learning algorithms. In the last step, a ground truth model was established by obtaining the actual tree numbers, and then the prediction performance was calculated by comparing the reference ground truth data with the proposed model. It is considered that significant success has been achieved for tree count with an average accuracy rate of 87% obtained using the MLP classifier in predetermined regions.