GNViT- An enhanced image-based groundnut pest classification using Vision Transformer (ViT) model

Area of Science:

• Agricultural Science
• Computer Science
• Artificial Intelligence

Background:

• Crop losses due to pests and diseases pose significant threats to global agriculture.
• Groundnut crops are particularly susceptible to pest-related damage, impacting food security.

Purpose of the Study:

• To introduce and evaluate the Groundnut Vision Transformer (GNViT) model for detecting and classifying groundnut pests.
• To assess the effectiveness of GNViT using standard reliability metrics and compare its performance against existing methods.

Main Methods:

• Utilized a pre-trained Vision Transformer (ViT) on the ImageNet dataset to develop the GNViT model.
• Trained and evaluated the model on the IP102 dataset, including pests like Thrips, Aphids, Armyworms, and Wireworms.
• Employed data augmentation techniques to enhance model training and performance.

Main Results:

• The GNViT model achieved a training accuracy of 99.52% after data augmentation.
• Evaluated model performance using F1-score, recall, and overall accuracy.
• Demonstrated superior accuracy compared to state-of-the-art methodologies in pest classification.

Conclusions:

• The GNViT model offers a reliable deep learning solution for classifying pests affecting groundnut crops.
• This technology has the potential to significantly reduce crop losses and contribute to global food security.
• Advanced AI solutions are crucial for addressing agricultural challenges and supporting a growing global population.