Apple Fruit Edge Detection Model Using a Rough Set and Convolutional Neural Network.

Area of Science:

• Computer Vision
• Machine Learning
• Agricultural Technology

Background:

• Accurate apple fruit detection is vital for intelligent picking and yield prediction.
• Existing edge detection methods struggle with complex agricultural environments.

Purpose of the Study:

• To develop an effective fruit edge detection algorithm for intelligent agriculture.
• To improve the accuracy and robustness of apple fruit detection.

Main Methods:

• Proposed a fusion edge detection model (RED) combining convolutional neural networks (CNN) and rough sets.
• Utilized Faster R-CNN for initial image segmentation and K-means clustering for noise reduction.
• Applied rough set theory to refine edge detection under varying illumination and occlusion.

Main Results:

• The RED model demonstrated high accuracy and robustness in detecting apple fruit edges.
• Significantly improved detection accuracy and stability compared to traditional operators.
• Performance was particularly notable under challenging conditions like complex backgrounds and varying illumination.

Conclusions:

• The RED model offers a promising solution for intelligent fruit picking and yield prediction.
• The fusion approach effectively addresses noise and environmental challenges in fruit detection.
• This research provides a foundation for advancing automated agricultural practices.