Penelitian ini bertujuan untuk mengevaluasi dan membandingkan performa algoritma YOLOv8 dan YOLO11 dalam dua fungsi utama, yaitu deteksi objek berbasis bounding box dan segmentasi objek berbasis mask, khususnya pada tubuh kepiting bakau (Scylla spp.). Hasil evaluasi menunjukkan bahwa YOLOv8 unggul dalam fungsi deteksi objek, dengan nilai precision, recall, accuracy, dan F1-score yang lebih tinggi dibandingkan YOLO11. Model ini juga menunjukkan kestabilan dalam pelokalan objek pada berbagai bentuk dan jumlah. Sebaliknya, pada fungsi segmentasi, YOLO11-Seg menunjukkan performa terbaik dengan nilai metrik evaluasi yang sangat tinggi, termasuk recall dan F1-score di atas 96%, serta kemampuan mengenali bagian tubuh kecil seperti kaki renang secara presisi. Dengan demikian, pemilihan model sebaiknya disesuaikan dengan kebutuhan implementasi; YOLOv8 cocok untuk sistem deteksi cepat dan efisien, sedangkan YOLO11-Seg lebih tepat digunakan pada aplikasi berbasis IoT yang memerlukan segmentasi visual detail.

This research was conducted to evaluate and compare the performance of the YOLOv8 and YOLO11 algorithms in two main tasks, namely object detection using bounding boxes and object segmentation using masks, applied to the body of mud crabs (Scylla spp.). Based on the evaluation results, YOLOv8 demonstrated better performance in object detection, as indicated by higher values of precision, recall, accuracy, and F1 score compared to YOLO11. In addition, YOLOv8 was found to produce more stable predictions and better object localization, particularly for complex shapes and larger object quantities. On the other hand, YOLO11-Seg achieved superior results in object segmentation, with recall and F1-score exceeding 96%, and was able to accurately segment small object parts such as swimming legs. It can be concluded that YOLOv8 is more suitable for fast and efficient detection systems, while YOLO11-Seg is recommended for IoT-based applications that require detailed and accurate pixel-level segmentation.