Related Experiment Video
Updated: Jun 3, 2025
10:25
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
Published on: November 11, 2022
8.5K
CabbageNet: Deep Learning for High-Precision Cabbage Segmentation in Complex Settings for Autonomous Harvesting
Yongqiang Tian1, Xinyu Cao1,2,3, Taihong Zhang1,2,3
1School of Computer and Information Engineering, Xinjiang Agricultural University, Urumqi 830052, China.
Sensors (Basel, Switzerland)
|January 8, 2025
Summary
This study introduces an enhanced YOLOv8n-seg network for precise real-time cabbage head segmentation in unmanned harvesting. The improved model significantly reduces damage and missed harvests, boosting agricultural efficiency.
Area of Science:
- Agricultural Robotics
- Computer Vision
- Machine Learning
Background:
- Unmanned harvesting requires accurate real-time segmentation of crop heads.
- Field conditions and crop variability pose challenges to precise segmentation.
- Reducing harvest damage and missed yields is crucial for efficiency.
Purpose of the Study:
- To develop an improved YOLOv8n-seg network for precise cabbage head segmentation.
- To enhance the accuracy and efficiency of automated cabbage harvesting systems.
- To address challenges posed by complex growing environments and cabbage morphology.
Main Methods:
- Modified the YOLOv8n-seg network's C2f module.
- Integrated deformable attention with dynamic sampling points.
- Introduced an ADown module to minimize detail loss during downsampling.
- Incorporated a Small Object Enhance Pyramid based on PAFPN for small target detection.
Main Results:
- Achieved Mask Precision of 92.2%, Mask Recall of 87.2%, and Mask mAP50 of 95.1%.
- Maintained a compact model size of 6.46 MB.
- Demonstrated superior accuracy and efficiency compared to mainstream instance segmentation models.
Conclusions:
- The proposed improved YOLOv8n-seg network effectively addresses challenges in segmenting field-grown cabbage.
- The model facilitates real-time, precise cabbage harvesting in complex environments.
- The enhancements contribute to reducing damage and missed harvest rates, improving overall efficiency.
