Related Experiment Video
Updated: May 8, 2025
14:15
On-Site Molecular Detection of Soil-Borne Phytopathogens Using a Portable Real-Time PCR System
Published on: February 23, 2018
16.4K
Potato late blight leaf detection in complex environments
Jingtao Li1, Jiawei Wu1, Rui Liu1
1Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming, 650504, China.
Scientific Reports
|December 27, 2024
Summary
An improved YOLOv5 algorithm enhances potato late blight detection. This lightweight model uses ShuffleNetV2 and attention mechanisms for higher accuracy in complex field conditions.
Area of Science:
- Agricultural Science
- Computer Vision
- Plant Pathology
Background:
- Potato late blight is a significant global crop disease.
- Accurate and efficient detection is crucial for disease management.
- Existing detection methods face challenges in complex environments.
Purpose of the Study:
- To develop an improved YOLOv5 algorithm for detecting potato late blight.
- To enhance detection accuracy and efficiency in complex agricultural settings.
- To create a lightweight and computationally efficient model.
Main Methods:
- Utilized ShuffleNetV2 as the backbone network for model optimization.
- Integrated a coordinate attention mechanism to improve detection of obscured or damaged leaves.
- Employed a bidirectional feature pyramid network for multi-scale feature fusion.
Main Results:
- Reduced model parameters from 7.02M to 3.87M and FLOPs from 15.94G to 8.4G.
- Increased detection speed by 16% and average precision by 3.22%.
- Achieved a more lightweight and efficient model with improved accuracy.
Conclusions:
- The enhanced YOLOv5 algorithm offers a robust solution for potato late blight detection.
- The model's efficiency and accuracy are suitable for complex environmental conditions.
- Findings support practical applications and further research in crop disease management.
