PLFYNet-Based Edge-Deployable Detection System for Ginkgo Biloba Leaf Diseases

Jun Wang¹Siyuan Gu¹Maocheng Zhao^2*

• ¹College of Information Science and Technology, Nanjing Forestry University, Nanjing, China
• ²College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing, China

Target detection is a pivotal technology for precise monitoring of leaf-used Ginkgo biloba diseases in precision agriculture. However, complex plantation environments impose significant constraints on existing detection systems, manifesting as degraded detection accuracy, suboptimal efficiency, and prohibitive computational overhead for edge deployment. This study proposes PLFYNet, a lightweight deep learning model specifically designed for real-time disease detection on resource-constrained embedded devices. First, we constructed a comprehensive multi-class dataset with 7,158 augmented images, encompassing three disease categories: chlorosis, insect pest, and physical damage. We systematically evaluated five lightweight architectures, adopted a reconstructed optimal backbone network, and integrated attention mechanisms, a detection head, and efficient convolution techniques to maintain architectural efficiency. We also developed a feature fusion module to address small target feature loss. After forming the model LCNet-FusionYOLO, we applied Layer-Adaptive Magnitude-based Pruning (LAMP) to reduce the model scale while enhancing its performance, ultimately yielding the PLFYNet model. The PLFYNet model attains 94.5% mAP@0.5 with only 3.0M parameters, surpassing YOLOv7-tiny by 4.8% while utilizing half the parameters. Deployment on Jetson Orin Nano demonstrates 50.5 FPS real-time inference, validating its practical applicability. This work establishes a paradigm for developing high-precision, computationally efficient disease detection systems, providing practical edge-based monitoring solutions for sustainable Ginkgo biloba cultivation.