Related Experiment Video
Updated: May 15, 2025
06:28
Author Spotlight: Unraveling Plant Responses to Abiotic Stresses Using the PlantScreen Robotic Platform
Published on: June 7, 2024
1.6K
Potato precision planter metering system based on improved YOLOv5n-ByteTrack
Cisen Xiao1, Changlin Song2, Junmin Li2
1School of Computer and Software Engineering, Xihua University, Chengdu, China.
Frontiers in Plant Science
|May 13, 2025
Summary
This study introduces an improved YOLOv5n model for real-time potato planting assessment, significantly reducing model size and computation while maintaining high accuracy. The system enables precise monitoring of planting operations, enhancing efficiency in potato cultivation.
Area of Science:
- Agricultural Engineering
- Computer Vision
- Machine Learning
Background:
- Manual assessment of potato planting is inefficient and lacks real-time evaluation capabilities.
- Current methods struggle to provide immediate feedback on planting accuracy and seed distribution.
Purpose of the Study:
- To develop an efficient and accurate detection algorithm for evaluating potato planting machine performance.
- To improve real-time monitoring of seed potato scooping and planting outcomes.
Main Methods:
- An improved lightweight YOLOv5n model incorporating C3-Faster and re-parameterized convolution (RepConv).
- Layer-adaptive magnitude-based pruning (LAMP) for model optimization on mobile devices.
- Integration with ByteTrack algorithm and a counting method for performance analysis.
Main Results:
- The enhanced YOLOv5n model achieved a 56.8% reduction in parameters, 56.1% decrease in GFLOPs, and 51.4% smaller model size.
- Mean average precision (mAP@0.5) reached 98.0%, with counting accuracy at 96.6%.
- The system provides real-time monitoring of omission, replanting, and qualified casting.
Conclusions:
- The proposed lightweight YOLOv5n model offers a significant improvement in efficiency and accuracy for potato planting assessment.
- The developed system supports precision potato planting with real-time visual feedback, demonstrating practical industrial value.
