Hybrid CNN-SVM Approach with Regularization for Accurate Classification of Images: A Case Study on Rudraksha Beads

Abstract

Based on the results obtained from the classification of the different types of Rudraksha beads using Based on the convolutional neural network model, it can be inferred that the model has attained a high level of precision in recognizing the various categories of beads. According to the evaluation metrics, the model has attained an accuracy rate of 67.55% overall. The F1-score, a metric that evaluates the model’s precision and recall, varies from 70.59% to 77.48%. The range of the support value, which reflects the number of occurrences in each category, is between 275 and 310. Additionally, the macro-average F1-score has been computed, which takes into account the performance of each class equally, is 75.46%. This is one of the research areas an agriculture. The weighted-average F1-score, which considers the importance of each class based on its support value, is 75.43%. The micro-average F1-score, which takes into account The combined number of accurate identifications, inaccurate identifications, and inaccurate negative identifications is 75.45%. Overall, the results demonstrate that the convolution neural network model is effective in identifying the different types of Rudraksha beads. The model has achieved high accuracy and F1 scores, indicating that it can be used for practical applications such as sorting and identifying the different types of beads. This is one of the research areas an agriculture.