JJEM: Special Edition 2 (December 2024)
JJEM: Special Edition 2 (December 2024)
2024-12-08
Valuable Stone Detection.
Akash V O, Sunitha G P
The study presents use of Convolutional Neural Network (CNN) to automate gemstone classification. The experiment starts by doing image preprocessing on the gemstones at a standard size of 256x256 pixels which means normalization has been achieved as well this is followed by augmentation through rotation plus flipping in order to improve model generalization. CNN consists of several convolutional layers with 2x2 filters activation function being ReLU this come together with another ones like dropout and batch normalization for regularization. The Adam optimizer with a learning rate of 0.0002 and Categorical Cross entropy loss function are used for training. Evaluation metrics like Accuracy, Mean Squared Error (MSE), and Top-3 Accuracy as well as diagnostic tools like confusion matrices and classification reports are employed in evaluation. Deep learning-based robust gemstone classification framework focuses on effective identification and categorization of gemstone varieties from the visual data in question, which is fundamental for gemological purposes.
Gemstone classification, CNN, diagnostic plots, model performance