Upgrading Coffee Bean Quality Using K-nearest Algorithm over Future Selection and Extraction to Reduce Dimensionality of Data
Chandu B, S Raveena, R Surendran · 2024
This research study explores the capacity of digital image processing and machine learning to assess the inherent quality of unprocessed coffee. The proposed approach includes the process of obtaining images, preparing them (which involves filtering and selecting relevant features), and analyzing them (which involves segmenting, extracting features, and classifying them). To categorize coffee quality according to morphological and color criteria, researchers used ANN, SVM, and K-Nearest Neighbors (KNN) classifiers. With an accuracy rate of 89.45% compared to 83.75% for the Support Vector Machine (SVM) algorithm and 77.85% for the k-Nearest Neighbors (KNN) strategy, the Artificial Neural Network (ANN) approach dominates the experimental results. To accurately evaluate the quality of unprocessed coffee, our results highlight the significance of careful image processing and classifier selection.