Optimising Corn Leaf Disease Classification with MobileNet and Oversampling

This work addresses the challenge of crop disease identification in agriculture by introducing an improved deep learning model. The study employs extensive data augmentation to enhance the training dataset, promoting better model generalisation across diverse agricultural scenarios. The model architecture utilises MobileNetV2 as a base, prioritising efficiency and lightweight design. The model's training progress is monitored over 40 epochs with a batch size of 10, resulting in an impressive accuracy of 93.3% in crop disease classification. The work also explores alternative preprocessing steps and assesses the impact of various base models for feature extraction. The outcomes contribute to advancing agricultural technology, providing reliable tools for crop management. These proposed enhancements offer valuable insights for researchers and practitioners in precision agriculture, facilitating sustainable and efficient practices. This work represents a significant step forward in applying deep learning techniques to achieve accurate and robust crop disease identification, with potential applications in improving crop yield, reducing losses, and promoting sustainable agricultural practices.