Innovative Approach to Detect Sugarcane Diseases with Discrete Fourier Transform and Deep Learning

Plant diseases threaten crop growth and production worldwide, threatening food security. Sugarcane, a vital global crop and primary source of sugar and ethanol, is particularly vulnerable to these diseases, which can devastate yields and cause financial hardships for small-scale farmers. This study tackles the problem by availing image processing with deep learning techniques. Before the training process, the input data set was refined using a Discrete Fourier Transformer to improve the primary data quality. A MobileNet model was trained by utilizing 6, 5 5 0 images of diseased and healthy sugarcane leaves and achieved a 9 3. 5 6 % accuracy. It effectively distinguishes between healthy and diseased leaves, providing farmers with a critical channel for early infection detection and management. This research emphasizes the transformative capabilities of deep learning models in aiding farmers by efficiently detecting and categorization sugarcane diseases.