darrayes/Kashmiri-Rice-Seeds

RiceNet Dataset: High-Quality Image Dataset for Rice Variety Classification

Overview

The RiceNet Dataset is a meticulously curated collection of rice grain images designed to facilitate the classification of five distinct rice varieties using deep learning techniques. This dataset underpins the RiceNet model, a Deep Convolutional Neural Network (DCNN) architecture developed to enhance the accuracy of rice variety identification, addressing challenges in agriculture and trade related to seed misclassification and grain adulteration.

Applications

• Agricultural Research: Assists in the development of automated systems for accurate rice variety classification, aiding in breeding programs and crop management.

• Trade and Quality Control: Supports the detection of grain adulteration, ensuring quality assurance in rice trade.

• Educational Purposes: Serves as a resource for academic projects and research in computer vision and agricultural informatics.

Performance Benchmarks

The RiceNet model, trained on this dataset, achieved a classification accuracy of 94%, outperforming traditional machine learning approaches:

• HOG-SVM: 66.0%

• SIFT-SVM: 65.33%

• HOG-LR: 62.67%

• SIFT-LR: 65.0%

• HOG-KNN: 54.0%

• SIFT-KNN: 52.0%

Additionally, pre-trained models like InceptionV3 and InceptionResNetV2 achieved accuracies of 84% and 81.33%, respectively.

Usage

Researchers and practitioners can utilize this dataset to train and evaluate models for rice variety classification. The dataset's structure is conducive to integration with popular deep learning frameworks such as TensorFlow and PyTorch.

Citation

If you use this dataset in your research, please cite the following paper:

Nusrat Mohi Ud Din, Assif Assad, Rayees Ahmad Dar, Muzafar Rasool, Saqib Ul Sabha, Tabasum Majeed, Zahir Ul Islam, Wahid Gulzar, Aamir Yaseen,
RiceNet: A deep convolutional neural network approach for classification of rice varieties,
Expert Systems with Applications,
Volume 235,
2024,
121214,
ISSN 0957-4174,
https://doi.org/10.1016/j.eswa.2023.121214.

Link to Paper

https://doi.org/10.1016/j.eswa.2023.121214

License

This dataset is available for public use under the [Creative Commons Attribution 4.0 International License].

Contact

For questions or collaborations, please contact Rayees Ahmad Dar at [email protected].

---