Improving Pea Variety Sorting with a Hybrid Approach of CNN-NB Method

This study introduces a new way to improve the reliability of pea variety identification by using two methods together: Convolutional Neural Networks (CNN) and Naive Bayes (NB). Types of peas have small differences in appearance, making it hard to sort them using old ways. In this study, the task was to classify different types of peas into three groups: Azad P-1, Pusa Pragati, and Arkel. The suggested plan combines the good things of CNN for getting features from picture data and NB for chance sorting. The CNN part finds complex patterns and space details in pea pictures, while the NB sorter improves sorting using chance modeling. The combination of these two ways tries to use their different good points, which makes a strong and true pea-kind identification system. Testing on a big group of data shows that the mixed model - CNN and NB together is good. The proposed model gets an overall accuracy of 96.2% in the case of the testing dataset. The study helps in how to look at farm pictures. It's useful for scientists and workers who work with peas, pea growing and improving them.