VotTomNet: Voting-based tomato disease diagnosis with transfer learning

Abstract

The research presents an advanced automation system, termed VotTomNet, designed for diagnosing tomato leaf diseases using transfer learning, and soft and hard voting ensemble techniques. By leveraging six pre-trained deep learning convolutional neural networks—VGG16, InceptionNet, ResNet, MobileNet, EfficientNet, and DenseNet—the system achieved an impressive accuracy of 99.2%. These models were meticulously fine-tuned to diagnose multiple types of tomato diseases with heightened precision. The integration of a soft and hard voting mechanism further enhanced the overall diagnostic accuracy by combining the strengths of these diverse models into a powerful ensemble. The findings underscore the robustness, reliability, and effectiveness of this ensemble technique, marking a significant advancement in precision agriculture and crop health assessment. By outperforming traditional methods, this approach offers a more practical and efficient solution for large-scale agricultural applications, enabling comprehensive crop management and improved yield. In conclusion, this research lays a strong foundation for future innovations in automated plant disease diagnosis and agricultural technology. Its contributions have the potential to revolutionize disease management, reduce crop losses, and ultimately enhance food security on a global scale.