Related Experiment Video
Updated: Jul 31, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
591
Grasping and cutting points detection method for the harvesting of dome-type planted pumpkin using transformer
Jin Yan1, Yong Liu1, Deshuai Zheng1
1School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China.
Frontiers in Plant Science
|May 5, 2023
Summary
This study introduces an advanced keypoint detection method for autonomous pumpkin harvesting. The novel framework enhances fruit and stem segmentation, improving grasping and cutting point accuracy for robotic systems.
Area of Science:
- Agricultural Robotics
- Computer Vision
- Machine Learning
Background:
- Accurate keypoint detection is crucial for autonomous harvesting systems.
- Existing methods struggle with overlapping instances in agricultural environments.
Purpose of the Study:
- To develop a robust keypoint detection method for autonomous pumpkin harvesting.
- To improve instance segmentation precision for fruits and stems in complex agricultural settings.
Main Methods:
- Proposed a pumpkin fruit and stem instance segmentation architecture fusing transformer and point rendering.
- Utilized a transformer network as the backbone for higher segmentation precision.
- Applied point rendering for finer masks, especially at boundaries of overlapping areas.
Main Results:
- Achieved mask mAP of 70.8% and box mAP of 72.0% for instance segmentation.
- Demonstrated significant gains (4.9% mask mAP, 2.5% box mAP) over Cascade Mask R-CNN.
- Validated the effectiveness of individual modules through ablation studies.
Conclusions:
- The proposed method offers a promising approach for keypoint estimation in fruit picking tasks.
- The framework enhances autonomous harvesting capabilities through improved segmentation and keypoint detection.
