Tea disease identification based on ECA attention mechanism ResNet50 network

Area of Science:

• Agricultural Science
• Computer Vision
• Plant Pathology

Background:

• Accurate identification of tea plant diseases is crucial for crop management.
• Complex backgrounds in tea gardens pose significant challenges for automated disease detection.
• Existing methods often struggle with feature extraction and background interference.

Purpose of the Study:

• To develop an advanced deep learning model for precise tea disease identification.
• To improve the accuracy and robustness of disease detection in complex field conditions.
• To enhance feature extraction and attention mechanisms for better target recognition.

Main Methods:

• Optimization of the ResNet50 architecture.
• Implementation of a multi-layer small convolution kernel strategy for enhanced feature extraction.
• Integration of the ECA (channel attention) mechanism to focus on critical disease features.

Main Results:

• The proposed ECA-ResNet50 model achieved an accuracy rate of 93.06% in tea disease identification.
• Demonstrated a 3.18% improvement in accuracy compared to the original ResNet50 model.
• Exhibited excellent generalization capabilities on datasets of other plant categories, mitigating complex background interference.

Conclusions:

• The ECA-ResNet50 model effectively addresses the challenge of tea disease identification in complex tea garden environments.
• The model's enhanced feature extraction and attention mechanisms lead to precise recognition of disease targets.
• ECA-ResNet50 shows significant potential for practical application in agricultural disease monitoring systems.