Using a novel convolutional neural network for plant pests detection and disease classification.

Area of Science:

• Agricultural Science
• Computer Science
• Artificial Intelligence

Background:

• Early detection of plant diseases and pests is vital to mitigate economic losses and reduce reliance on harmful chemicals.
• Threats to global food security arise from inefficient agricultural practices and pest management.

Purpose of the Study:

• To develop an advanced model for accurate plant pest and disease identification and classification.
• To improve the efficiency and effectiveness of agricultural monitoring systems.

Main Methods:

• An enhanced Convolutional Neural Network (CNN) integrated with Long Short-Term Memory (LSTM) was proposed.
• A majority voting ensemble classifier was employed for robust classification.
• Deep features were extracted from pre-trained models and fed into the LSTM layer.
• Experiments were conducted on a dataset of 4447 apple pests and diseases across 15 classes.

Main Results:

• The proposed LSTM-CNN model achieved an accuracy of 99.2%.
• Performance was validated against Logistic Regression (LR) and Extreme Learning Machine (ELM) classifiers.
• The model outperformed existing transfer learning techniques.

Conclusions:

• The developed model demonstrates superior performance in plant pest and disease detection.
• The integration of CNN and LSTM with ensemble methods offers a powerful solution for agricultural monitoring.
• Further research can explore the application of specific pre-trained model layers for enhanced accuracy.