Revolutionizing Gemstone Classification with Hybrid CNN-KNN Models

This paper thoroughly analyzes the classification of gemstones using a hybrid CNN-KNN model. It shows high precision and reliability for Onyx Red, Pyrope, Rhodolite, and Zoisite. The model achieved impressive precision values: Onyx Red at 95.94%, Pyrope at 97.43%, Rhodolite at 96.07%, and Zoisite at 97.45%. Recall values were also high, with Onyx Red at 97.38%, Pyrope at 96.01%, Rhodolite at 97.46%, and Zoisite at 96.06%. Corresponding F1 scores were 96.65% for Onyx Red, 96.72% for Pyrope, 96.76% for Rhodolite, and 96.75% for Zoisite, indicating a balanced performance in precision and recall. The dataset included 2670 samples of Onyx Red (24% support proportion), 2760 samples of Pyrope (25% support proportion), 2760 samples of Rhodolite (25% support proportion), and 2790 samples of Zoisite (25% support proportion). The model's total accuracy was 96.72%, having been trained on a dataset of 10,980 pictures. The model's accuracy for each class was uniformly high at 98%, highlighting its effectiveness in accurate gemstone identification. These metrics validate the robustness of the hybrid CNN-KNN approach, showcasing its potential to significantly enhance the efficiency and precision of gemstone classification. The study emphasizes the potential of combining convolutional neural networks for feature extraction with k-nearest neighbors for classification, leveraging the strengths of both methods. The high-performance metrics across all classes indicate that the model can reliably classify gemstones, reducing reliance on human expertise and minimizing errors. This research offers promising implications for the gemstone industry, suggesting that advanced AI techniques can revolutionize gemstone identification by providing automated, accurate, and efficient classification solutions.