LBS-YOLO: a lightweight model for strawberry ripeness detection
Haitao Fu1, Xueying Li1, Zheng Li1
1College of Information Technology, Jilin Agricultural University, Changchun, China.
Frontiers in Plant Science
|December 22, 2025
Summary
This study introduces LBS-YOLO, a lightweight model for strawberry picking. It enhances recognition accuracy and efficiency, addressing limitations in current intelligent picking technologies.
Area of Science:
- Computer Vision
- Robotics
- Agricultural Technology
Background:
- Traditional strawberry picking relies on manual labor, facing challenges with an aging population.
- Existing intelligent picking technologies suffer from suboptimal recognition accuracy and low computational efficiency.
Purpose of the Study:
- To develop a lightweight and efficient strawberry detection model for intelligent picking.
- To improve recognition accuracy and reduce computational load compared to existing methods.
Main Methods:
- Constructed a lightweight detection model, LBS-YOLO, based on an improved YOLOv11n architecture.
- Incorporated a lightweight LAWDS module for enhanced feature representation and a Bidirectional Feature Pyramid Network (BiFPN) for effective feature fusion.
- Utilized a C3k2_Star module to replace the conventional C3K2 for superior feature representation.
Main Results:
- LBS-YOLO achieved 88.6% mAP@0.5 and 75.8% mAP@0.5:0.95, outperforming YOLOv11n by 2.2% and 1.3% respectively.
- Improved recall rate to 86.4% and F1-score to 82.9%, with a reasoning speed of 260.7 FPS.
- Reduced storage space by 34.6% and parameter count by 38% compared to YOLOv11n.
Conclusions:
- LBS-YOLO significantly reduces parameters while enhancing detection accuracy and operational efficiency.
- The model effectively mitigates false and missed detections, offering robust support for automated strawberry harvesting and monitoring.
