DeepRiceTransfer: Exploiting CNN Transfer Learning for Effective Rice Variety Classification

Abstract- The proposed DeepRiceTransfer (DRT) model gives a thorough examination of the accurate categorization of rice types. The research aimed at reworking the accuracy and effectiveness of rice variety types. Convolutional Neural Networks (CNNs) and transfer learning are utilized by the version to strategically capitalize on previous facts from several datasets. The proposed methodology used many extraordinary neural network architectures, along with the DeepRiceTransfer Convolutional Neural community (CNN) for the rice image dataset, to achieve the correct type of results. To enhance the model’s overall performance, we especially used transfer learning methods with the MobileNet, ResNet50, and VGG16 architectures. Numerous statistical criteria, which include accuracy, precision, F1 score, and support, have been used to evaluate the class consequences. With accuracy scores of 98.44% for the DRT model,98.94% for CNN with transfer learning using MobileNet, 79.79% for ResNet50, and 99.47% for VGG16 architectures, the models showed impressive success in classifying rice varieties. The models confirmed impressive achievement in classifying rice types effects spotlight CNN models’ effectiveness in efficaciously classifying rice sorts consistent with their distinctive trends, especially those who use DRT and switch getting to know with MobileNet, ResNet50, and VGG16 architectures. The models’ elevated accuracy charges spotlight their capacity for real-time implementations in the rice range category and first-class assessment.