Design of Citrus Fruit Detection System Based on Mobile Platform and Edge Computer Device

Area of Science:

• Agricultural Engineering
• Computer Vision
• Artificial Intelligence

Background:

• Accurate citrus fruit detection is crucial for orchard management and yield estimation.
• Challenges in mountainous orchards include leaf and fruit occlusion, hindering traditional methods.

Purpose of the Study:

• To develop an efficient and accurate citrus fruit detection system for mountainous orchards.
• To improve existing target detection algorithms for enhanced performance in complex environments.

Main Methods:

• Utilized unmanned aerial vehicle (UAV) data collection for comprehensive orchard imagery.
• Enhanced a state-of-the-art target detection algorithm with attention mechanisms and adaptive feature fusion.
• Applied model pruning for efficient deployment on edge computing devices.

Main Results:

• Achieved a citrus detection accuracy of 93.32% on a custom dataset.
• Demonstrated a processing speed of 180 ms/frame on the edge computing device.
• The improved algorithm effectively handles occlusion challenges in orchard environments.

Conclusions:

• The proposed system offers a viable solution for citrus fruit detection in challenging mountainous orchards.
• This technology supports precise orchard management and accurate yield determination for growers.