Related Experiment Video
Updated: Jun 16, 2025
04:21
Author Spotlight: Sieving Fruit Pulp to Detect Immature Tephritid Fruit Flies in the Field
Published on: July 28, 2023
1.7K
ELD-YOLO: A Lightweight Framework for Detecting Occluded Mandarin Fruits in Plant Research
Xianyao Wang1, Yutong Huang1, Siyu Wei2
1College of Information Engineering, Sichuan Agriculture University, Ya'an 625014, China.
Plants (Basel, Switzerland)
|June 13, 2025
Summary
ELD-YOLO improves mandarin fruit detection in orchards by enhancing edge details and handling occluded fruits. This lightweight model offers higher precision and recall while reducing computational costs for efficient harvesting.
Area of Science:
- Computer Vision
- Agricultural Technology
- Machine Learning
Background:
- Mandarin fruit detection is vital for yield prediction and harvesting.
- Complex orchard environments present challenges like fruit occlusion and small fruit visibility.
- Existing models struggle with efficiency and accuracy in these scenarios.
Purpose of the Study:
- To develop a lightweight detection framework (ELD-YOLO) for enhanced small and occluded mandarin fruit detection.
- To improve edge detail preservation and model efficiency in complex orchard settings.
Main Methods:
- Proposed ELD-YOLO, a lightweight detection framework.
- Incorporated edge-aware processing for strengthened feature representation.
- Introduced a streamlined detection head and adaptive upsampling strategy.
Main Results:
- Achieved 89.7% precision, 83.7% recall, 92.1% mAP@50, and 68.6% mAP@50:95.
- Reduced parameter count by 15.4% compared to the baseline model.
- Demonstrated superior performance in detecting small and occluded fruits.
Conclusions:
- ELD-YOLO offers an effective and efficient solution for mandarin fruit detection in challenging orchard environments.
- The framework balances high accuracy with reduced computational cost.
- Enables precise fruit identification and harvesting support.
