FusedNet Model for Varietal Classification of Rice Seeds

Rice is one of the most cultivated crops in the world and a primary food source for more than half of the global population. The primary focus of this paper is to implement an ensemble learning model, i.e., FusedNet, that aims to precisely classify 90 different rice seed varieties by utilizing both Red-Green-Blue (RGB) and hyperspectral images (HSI). The FusedNet model comprises two classifiers: first, the support vector machine (SVM) classifier that utilizes spatial and spectral features extracted from the RGB and hyperspectral image data, respectively, and second, the ResNet-50 network (based on Convolutional Neural Network) trained using single seed RGB image data. The model proposed in this study achieved an impressive testing accuracy score of 87.27% and an average F1-score of 86.87%, surpassing the results of the prior investigation conducted on the same publicly available dataset.