Advancing Mango Leaf Disease Detection with YOLOv11 Incorporating White Scale and Cottony Cushion Scale

Mango cultivation plays a vital role in India’s agricultural economy, but leaf diseases such as Bacterial Black Spot, Die Back, Sooty Mould, White Scale, and Cottony Cushion Scale cause significant yield losses. Among these, White Scale and Cottony Cushion Scale are rarely studied in the existing literature, and no prior work has addressed application of deep learning approaches for mango leaf disease detection. Traditional diagnosis approaches are slow, error-prone, and unsuitable for large-scale monitoring. This paper proposes a novel YOLOv11-based deep learning framework for real-time mango leaf disease detection on a custom 3,000-image dataset collected from Saharanpur, India, covering five major diseases and healthy leaves. The proposed model was benchmarked against recent state-of-the-art (SOTA) approaches, including ResNet50, EfficientNetB0, MobileNetV3, and DenseNet78. YOLOv11 achieved a precision of 0.964, a recall of 0.976, an F1-score of 0.97, and a mAP@0.5 of 0.983, outperforming ResNet50 by +2.4% and EfficientNetB0 by +1.1% in mAP. A non-parametric Wilcoxon signed-rank test confirmed the statistical significance (p < 0.05) of the improvements over all baselines. Furthermore, the model was integrated into an Android application for real-time deployment, enabling early detection and treatment to support precision agriculture. This contribution demonstrates not only improved accuracy but also the first inclusion of White Scale and Cottony Cushion Scale detection using YOLOv11, making it a novel step forward for mango disease management and sustainable farming.