Manuscript received March 6, 2025; revised April 30, 2025; accepted June 26, 2025; published September 18, 2025.

Abstract—Image segmentation is a crucial task in computer vision techniques, serving as a fundamental method for partitioning images into detailed segments that facilitate analysis and retrieval. This paper examines the performance of three YOLO models—particularly YOLOv5, YOLOv8, and YOLOv11—in nature image segmentation, specifically focusing on reptile images. The experiment evaluates accuracy, precision, recall, mean Average Precision (mAP), and computational efficiency using a diverse dataset of reptile images captured under varying environmental conditions. In the conducted experiments, we performed comparative tests involving the three models, yielding distinct outputs. Each of these models has its advantages, highlighting the best performance traits of each. YOLOv5 is user-friendly in implementation, YOLOv8 operates effectively without anchors, and YOLOv11 exhibits greater efficiency compared to the other two models. The results indicate that YOLOv11 has made significant advancements in architecture and training methods, establishing it as a versatile option for a range of computer vision tasks.

Keywords—computer vision, yolo, image segmentation, nature detection

Cite: Timothy Harminto, Dimas F. Najib, Pandu Wicaksono, and Zahra Nabila Izdihar, "Performance Comparison of YOLOv8, YOLOv5 and YOLOv11 in Nature Image Segmentation," Journal of Image and Graphics, Vol. 13, No. 5, pp. 515-520, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.