Banana Leaf Disease Detection using Vision Transformer model

The Banana Leaf Disease Detection project provides an automated and accurate solution for detecting diseases in banana leaves using a hybrid of Vision Transformer and CNN models. This approach reduces manual inspection, enhances early detection, and improves crop management.

• The hybrid Vision Transformer and CNN model achieved significant improvements in the detection of banana leaf diseases, ensuring reliable results.
• The system enables early detection of diseases such as cordana, sigatoka, and pestalotiopsis, helping farmers to take proactive measures.
• By automating disease detection, the system minimizes the need for manual inspections, saving time and labor costs for farmers.
• Early and accurate disease detection allows for better disease management, leading to improved crop yields and healthier banana plants.
• The application of class weights and data augmentation ensures that the models maintain balanced performance across all disease categories, avoiding bias toward dominant classes.
• The system’s precision and recall scores were optimized, ensuring that diseases were detected with minimal false positives and false negatives.
• The integration of predictions from both Vision Transformer and CNN models through a voting mechanism improved the overall accuracy of the system.
• The model architecture is designed to be scalable and adaptable, allowing for future application in detecting other types of plant diseases.
• By identifying diseases early, farmers can prevent the spread of infection, reducing crop losses and safeguarding their income.
• The automation of disease detection significantly reduces both the time and cost associated with traditional inspection methods, making it an economical solution for large-scale farming.