# Automatic Detection and Counting of Coconut Fruits in Overlapping Conditions on Trees Using Drones > Andri A. URL kanonis: https://discover.unhas.ac.id/publications/pub_scopus_105055437861 Jurnal / Konferensi: 2026 IEEE International Conference on Artificial Intelligence and Mechatronics Systems Aims 2026 Tahun terbit: 2026 DOI: https://doi.org/10.1109/AIMS71270.2026.11712750 Citations: 0 ## Authors - Andri A. ## Abstract Indonesia is one of the world’s largest coconut producers; however, the efficiency of the pre-harvest phase remains constrained by conventional manual maturity assessment methods. Visual challenges, including tree heights of 10–30 meters, frond occlusion, canopy shadows, and overlapping fruits, reduce the accuracy of production estimation. This study proposes an automated coconut maturity detection, segmentation, and counting system using Unmanned Aerial Vehicle (UAV) imagery based on a Deep Learning approach. To address illumination variations and occlusion problems, image preprocessing techniques based on Gamma Correction and Contrast Limited Adaptive Histogram Equalization (CLAHE) were applied prior to training the YOLOv11L-seg model. Experimental results on image-based object detection demonstrated that the proposed preprocessing framework improved the mean Average Precision (mAP@0.5) by 4.00%, increasing performance from the baseline model of 78.6% to 82.6%, with AP scores of 85.5% for the Young Coconut class and 79.7% for the Old Coconut class. The effectiveness of the proposed framework was further validated through UAV video-based automated counting experiments, where the baseline system without preprocessing achieved a counting accuracy of only 69.35% due to occlusion and uneven illumination. In contrast, the integration of Gamma-CLAHE significantly improved the counting accuracy to 82.25%. These results demonstrate that the proposed framework is robust and adaptive for handling complex plantation environments, making it suitable for precision agriculture applications. ## Keywords - Drone - Artificial intelligence - Computer science - Mathematics - Tree (set theory) - Noise (video) - Pattern recognition (psychology) - Biology - Environmental science - Measure (data warehouse) --- Sumber: Discover Unhas — RIMS Universitas Hasanuddin. Saat mengutip, gunakan DOI bila tersedia atau URL kanonis di atas.