Using image augmentation techniques and convolutional neural networks to identify insect infestations on tomatoes.

Area of Science:

• Agricultural Science
• Computer Science
• Artificial Intelligence

Background:

• Insect pests pose a significant threat to regional and global food security, particularly impacting vital crops like tomatoes in Jordan.
• Effective pest identification is crucial for timely intervention and crop protection to prevent agricultural losses.

Purpose of the Study:

• To develop and evaluate a deep learning model for accurate identification of insect pests affecting tomato crops in Jordan.
• To enhance pest detection capabilities for farmers through accessible mobile applications.

Main Methods:

• A dataset of insect pest images was curated, augmented to 5894 images, and split into 80% training and 20% validation sets.
• Convolutional Neural Networks (CNNs) were employed as the deep learning model for image classification.
• Image augmentation techniques were utilized to increase dataset size and improve model robustness.

Main Results:

• The trained CNN model achieved 90% training accuracy, 85% testing accuracy, and 87% validation accuracy.
• The developed model demonstrates significantly higher accuracy compared to previous methods (50-60%) without image augmentation.
• The study highlights the effectiveness of deep learning and image augmentation in pest identification.

Conclusions:

• A high-accuracy deep learning model for insect pest detection was successfully developed, offering a valuable tool for farmers.
• The model's potential for deployment on mobile applications can provide real-time pest identification support.
• This research contributes to improving agricultural practices and safeguarding food security through advanced technology.