Automated detection and counting of palm trees is a significant area of research for numerous countries, including the UAE. Currently, all palm trees are counted and monitored manually, a labor-intensive process that demands significant time and effort. However, the UAE has recently made significant advancements in remote sensing technologies, creating an opportunity to integrate space technology with agriculture for the efficient monitoring of palm trees throughout the country. This research paper presents a novel High-Resolution (HR) remote sensing dataset designed for the autonomous detection of palm trees in the UAE. The dataset has been acquired using Unmanned Aerial Vehicles (UAVs) covering various areas within UAE, including Ajman, Dubai, Khorfakkan, and Al-Ain. This paper utilizes the introduced dataset to evaluate the strengths and weaknesses of four object detection neural networks; You Only Look Once (YOLO)-v4 and -v5, Faster Region-based Convolutional Neural Network (FRCNN), and Detection Transformer (DETR). YOLOv5s achieved outstanding performance, with an Average Precision (AP) of 96.6% and Average F1- score (AF) of 95.7%, demonstrating its effectiveness in accurately detecting and localizing palm trees. Moreover, model outputs are effectively integrated into Geographic Information Systems (GIS) for enhanced spatial analysis and monitoring.