2025/12/15 by S. Sasitharan, Sridevi Subbiah, Nirmala Devi Malaichamy +1 · 1 voice
Agricultural and Biological Sciences · Environmental Science · #Smart Agriculture and AI #Remote Sensing in Agriculture #Plant Disease Management Techniques
paper · pdf · doi:10.70389/pjs.100164
openalex publication_date 2025/12/15 · openalex created_date 2025/12/17 · openalex updated_date 2026/06/11
Early detection of mango leaf diseases is essential for ensuring healthy crop yields and preventing economic losses. Traditional methods struggle to balance accuracy and efficiency, especially in resource-limited settings. This study explores the effectiveness of MobileNetV2, a lightweight deep learning model, for mango leaf disease detection. On a balanced multi-class dataset, MobileNetV2 achieves 93.4% accuracy (mean ± 0.5 SD across 5 folds) on the test set after 32 epochs. The model showed efficiency and robustness, with precision 0.92, recall 0.91, and F1-score 0.91 on the test set.The novelty of this work lies in three aspects: (i) a balanced dataset with a robust preprocessing and augmentation pipeline, (ii) measured real-time deployment on a Raspberry Pi 4 with inference latency, throughput, and memory usage reported, and (iii) integration of field-level insights with directions for attention-based and ensemble enhancements. These contributions differentiate our work from prior MobileNetV2 studies that relied on controlled datasets or theoretical Floating Point Operations (FLOPs). Overall, MobileNetV2 provides a lightweight, scalable, and cost-effective solution for real-time mango leaf disease detection, contributing to improved precision agriculture and sustainable farming practices.