Description of Invention
CASSAVA-SPECTRA2D is a novel approach that enhances early cassava disease classification by converting 1D spectral data into 2D images. Early detection of plant diseases like Cassava Mosaic Disease (CMD) and Cassava Brown Streak Disease (CBSD) is challenging due to subtle early-stage symptoms. While spectroscopy is effective for diagnosing asymptomatic conditions, interpreting raw 1D spectral data is complex due to overlapping patterns. This study compares traditional machine learning methods with a 1D-CNN for direct 1D spectral classification and pretrained CNN models for 2D image classification. Various padding strategies are also evaluated. Our findings demonstrate that this approach, combined with effective padding techniques, significantly enhances classification accuracy when converting 1D data into 2D images compared to direct 1D data classification.