Search

The palm sector plays a critical role in achieving the Sustainable Development Goals (SDGs). Accurate identification and counting of palm trees are vital for estimating yields, managing orchards, and planning future planting. We are currently investigating deep-learning-based object detection algorithms and remote sensing (RS) technologies to enhance industry efficiency. This study aimed to develop an efficient method for detecting and locating individual palm trees using unmanned aerial vehicles (UAVs) equipped with RS technology and deep learning algorithms.We conducted a comparative evaluation of four variations of YOLOv5 (YOLOv5-6l, YOLOv5-L, and YOLOv5-L-HighAug) and one variation of YOLOv8 (YOLOv8-HighAug). These models were trained, validated, and tested on a dataset of palm tree images. The results show that YOLOv8-HighAug consistently outperformed the other models, achieving an Average Precision (AP) of 0.88, a precision of 0.87, and a recall of 0.86. YOLOv5-L-HighAug also exhibited high AP@50 values (0.83), making it suitable for precise detection tasks while balancing resource efficiency.The proposed study reveals the potential of integrating deep learning algorithms with UAVs and RS to revolutionize practices in the palm industry. By reducing human errors and increasing data collection speed, this approach offers significant benefits for accurate detection and management of palm trees in agricultural landscapes. In conclusion, embracing technology such as deep learning-based object detection can improve efficiency and sustainability within the palm industry. Implementing modern palm tree identification and mapping techniques can lead to improved decision-making, resource optimization, and successful achievement of SDGs in the palm sector.