A Hybrid DL with Battle Royal Optimisation Algorithm for Accurate Tree Counting Using Satellite Images

Tree enumeration is a fundamental task in environmental monitoring, sustainable forestry management, and urban planning, yet manual methods remain prohibitively time-consuming and labor-intensive. This study presents an innovative approach, named BRO (Briefly Optimized Recognition with Deep Learning (DL) for accurate and efficient tree enumeration utilizing high-resolution satellite imagery and advanced machine learning techniques, specifically leveraging DL and transfer learning for robust tree detection and counting in complex environments. Experimental results demonstrate the significant effectiveness of the BRO approach compared to baseline methods, achieving a high accuracy of 97.8%. Furthermore, BRO shows substantial improvements in counting precision, resulting in a 5% reduction in Root Mean Squared Error (RMSE) and a 7% decrease in Mean Absolute Error (MAE) over existing techniques. Beyond performance metrics, execution time benchmarks highlight BRO’s computational efficiency, processing large datasets significantly faster than conventional optimization methods, which is crucial for large-scale applications. This research provides a robust and efficient system critical for various real-world applications, including large-scale deforestation monitoring, afforestation project planning and evaluation, and detailed urban forest inventories, thereby facilitating informed decision-making for environmental conservation and resource management.