Related Experiment Video
Updated: Jan 10, 2026
08:47
Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
Published on: February 9, 2024
2.0K
MobileNet-GDR: a lightweight algorithm for grape leaf disease identification based on improved MobileNetV4-small
Gang Chen1, Zhennan Xia1, Xiaodan Ma1
1School of Information Engineering, Changchun College Of Electronic Technology, Changchun, China.
Frontiers in Plant Science
|November 24, 2025
Summary
This study introduces MobileNet-GDR, a lightweight deep learning model for grape leaf disease recognition on mobile devices. It achieves high accuracy with reduced computational cost, enabling real-time field diagnosis.
Area of Science:
- Agricultural technology
- Computer vision
- Deep learning
Background:
- Deep learning models for grape leaf disease diagnosis face challenges with computational complexity and mobile deployment.
- Existing models often require significant resources, limiting their use in field conditions.
Purpose of the Study:
- To develop a lightweight image classification algorithm for efficient grape leaf disease diagnosis on mobile devices.
- To optimize deep learning models for real-time performance and accuracy in agricultural applications.
Main Methods:
- Proposed MobileNet-GDR algorithm based on MobileNetV4-small architecture.
- Utilized depthwise separable convolutions and grouped convolutions for efficient feature extraction and fusion.
- Incorporated PReLU activation functions to improve nonlinear representation.
Main Results:
- MobileNet-GDR achieved 99.625% classification accuracy with only 1.75M parameters and 0.18G FLOPs.
- Real-time inference speed of 184.89 FPS was attained.
- Demonstrated superior computational efficiency compared to FasterNet and GhostNet.
Conclusions:
- MobileNet-GDR offers a practical and lightweight solution for real-time grape leaf disease diagnosis.
- The model's efficiency and accuracy are valuable for agricultural applications and field disease monitoring.
