Comparative performance of next-Gen YOLO models for leaf health classification in ornamental species

Automated plant disease detection has become an essential application of deep learning, supporting early diagnosis and effective crops and ornamental plant management. Recent advancements in the You Only Look Once (YOLO) family of object detection models have improved both accuracy and efficiency, making them suitable for real-time deployment. This paper presents a comparative analysis of YOLOv8, YOLOv9, and YOLOv11 for classifying diseased and healthy leaves of Ixora and Bougainvillea, two widely grown ornamental species. A curated dataset of annotated leaf images covering multiple disease conditions was used to train and evaluate the models under consistent experimental settings. To capture both accuracy and real-time feasibility, performance was evaluated using standard detection metrics like mean Average Precision (mAP), precision, recall, and F1-score in addition to inference speed (FPS). The assessment also highlights environmental robustness and subtle disease localization parameters, which are important for monitoring ornamental plants in unrestricted outdoor environments. Results indicate that YOLOv11 achieves the highest detection accuracy, especially in capturing subtle disease patterns, while YOLOv8 and YOLOv9 demonstrate competitive performance with faster inference, making them preferable for resource-limited applications. The findings highlight practical trade-offs between accuracy and efficiency across YOLO versions, offering valuable insights for real-world deployment. By extending research beyond staple crops to ornamental plants, this work underscores the broader applicability of AI-driven disease detection and establishes a benchmark for evaluating next-generation YOLO architectures in horticulture.