Beyond 2D: 3D Point Cloud Analysis with Color Extraction and Precise Percentage Calculation of Anthracnose in Chillies

This paper introduces a novel Deep Learning approach leveraging the ResNet50 Architecture for the classification of diseased red chili peppers. Beyond conventional 2D methods, the proposed methodology incorporates a 3D point cloud approach to accurately determine the percentage of rotten chili, aiming to revolutionize crop health management. The comprehensive process includes data collection, 2D image classification, 2D to 3D point cloud conversion, and defected region extraction, applied to a meticulously assembled dataset. ResNet-50 proves effective, and the integration of a color exclusion algorithm enhances the analysis of diseased regions, providing a quantitative metric for evaluating filtering processes. This methodology significantly contributes to the resilience of chili crop production and holds promise for broader applications in plant disease detection, emphasizing the importance of integrating 3D point cloud methodologies in agricultural research and technology.