Incremental RPN: Hierarchical Region Proposal Network for Apple Leaf Disease Detection in Natural Environments

Area of Science:

• Agricultural Science
• Computer Vision
• Plant Pathology

Background:

• Apple leaf diseases significantly impact crop yield and quality.
• Accurate disease detection is vital for efficient agricultural monitoring.
• Challenges exist in distinguishing leaf lesions from background noise in natural settings.

Purpose of the Study:

• To develop an accurate method for detecting apple leaf diseases in natural environments.
• To address the challenge of lesion confusion with background noise.
• To improve the efficiency of apple disease monitoring systems.

Main Methods:

• A novel cascaded Incremental Region Proposal Network (Inc-RPN) was proposed.
• The Inc-RPN features a two-layer architecture: a precursor RPN for diseased leaf proposals and a successor RPN for disease spot extraction.
• Key components include a low-level feature aggregation module, an incremental module, and a novel position anchor generator.

Main Results:

• The proposed Inc-RPN demonstrated high performance on the FALD_CED and Apple Leaf Disease datasets.
• The method successfully achieved accurate detection of apple leaf diseases.
• The cascaded architecture and feature aggregation modules proved effective.

Conclusions:

• The Inc-RPN is a highly effective deep learning model for apple leaf disease detection.
• The proposed network overcomes challenges posed by complex natural environments.
• This technology can significantly enhance agricultural monitoring and disease management.