Research Article

Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models

Volume: 29 Number: 1 April 25, 2025
TR EN

Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models

Abstract

Abstract: Achieving high accuracy rates in the field of image processing often exceeds the limits of a single model. Therefore, hybridizing XGBoost and deep learning models is a common approach to obtaining more accurate and reliable results. Deep learning models are highly capable of extracting complex and meaningful features from images. However, to effectively classify these features, the use of a powerful machine learning algorithm like XGBoost can further enhance performance. Hybrid models combine the best features of both models, allowing them to achieve higher accuracy rates that would not be possible if used individually. High accuracy improves the model's reliability and effectiveness in application, thereby preventing misclassification and improving overall performance. Therefore, hybridization of models is essential for better results. In this paper, after flattening the extracted features, an XGBoost-based model was trained by utilizing decision trees, and the model achieved an accuracy of 98.813% on the test data. SHAP and XAI LIME were employed to explain the model, providing visualizations of how the features impacted the model's decisions positively or negatively based on their weight values, and demonstrating how these features influenced the decision-making process.

Keywords

References

  1. [1] Food Losses and Waste (FAO) Inventory and management at each stage in the food chain, 2016. https://openknowledge.fao.org/server/api/core/bitstreams/36cb45bc-392c-41fb-97f1-90ca1f16ee7f/content (Access date: 25.12.2024)
  2. [2] Bhattiprolu, S. 2021. XGBoost for image classification using VGG16. https://github.com/bnsreenu/python_for_microscopists/blob/master/195_xgboost_for_image_classification_using_VGG16.py (Access date: 26.03.2024)
  3. [3] Kaggle, 2018, Fruits Fresh and Rotten for Classification, https://www.kaggle.com/datasets/sriramr/fruits-fresh-and-rotten-for-classification (Access date: 26.02.2024)
  4. [4] Mendeley Data, 2022, Fresh and Rotten Fruits Dataset for Machine-Based Evaluation of Fruit Quality, https://data.mendeley.com/datasets/bdd69gyhv8/1 (Access date: 25.12.2024)
  5. [5] Mehedi, H., Moinul, H. 2021. Fresh and Rotten Fruit Classification Using Deep Learning. Daffodil International University. Computer Science and Engineering.
  6. [6] Ayşin, B. 2022. Determining the Freshness of Fruits with Deep Learning Methods. Tekirdağ Namık Kemal Üniversitesi. Bilgisayar Mühendisliği Anabilim Dalı. Yüksek Lisans Tezi. Tekirdağ.
  7. [7] Hithesh Kumar, C.M., Vikash, VArun, Dyamanagouda, P. 2021. Detection of Quality of Fruits Usıng AI. International Research Journal of Modernization in Engineering Technology and Science, 3(7).
  8. [8] Sohel, M., Tayeeba, T., Mirajul., I., Mumenunnesa, K., Riazue, R., Syed, A.H. 2021. An Advanced Method of Identification Fresh and Rotten Fruits using Different Convolutional Neural Networks. International Conference on Computing Communication and Networking Technologies (ICCCNT) 12th

Details

Primary Language

English

Subjects

Quantum Engineering Systems (Incl. Computing and Communications)

Journal Section

Research Article

Publication Date

April 25, 2025

Submission Date

October 30, 2024

Acceptance Date

March 12, 2025

Published in Issue

Year 2025 Volume: 29 Number: 1

APA
Şengöz, N., Köroğlu, H., & Kırıktaş, B. N. (2025). Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models. Süleyman Demirel Üniversitesi Fen Bilimleri Enstitüsü Dergisi, 29(1), 124-133. https://doi.org/10.19113/sdufenbed.1575098
AMA
1.Şengöz N, Köroğlu H, Kırıktaş BN. Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models. J. Nat. Appl. Sci. 2025;29(1):124-133. doi:10.19113/sdufenbed.1575098
Chicago
Şengöz, Nilgün, Harun Köroğlu, and Beyza Nur Kırıktaş. 2025. “Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm With Explainable Artificial Intelligence Models”. Süleyman Demirel Üniversitesi Fen Bilimleri Enstitüsü Dergisi 29 (1): 124-33. https://doi.org/10.19113/sdufenbed.1575098.
EndNote
Şengöz N, Köroğlu H, Kırıktaş BN (April 1, 2025) Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models. Süleyman Demirel Üniversitesi Fen Bilimleri Enstitüsü Dergisi 29 1 124–133.
IEEE
[1]N. Şengöz, H. Köroğlu, and B. N. Kırıktaş, “Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models”, J. Nat. Appl. Sci., vol. 29, no. 1, pp. 124–133, Apr. 2025, doi: 10.19113/sdufenbed.1575098.
ISNAD
Şengöz, Nilgün - Köroğlu, Harun - Kırıktaş, Beyza Nur. “Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm With Explainable Artificial Intelligence Models”. Süleyman Demirel Üniversitesi Fen Bilimleri Enstitüsü Dergisi 29/1 (April 1, 2025): 124-133. https://doi.org/10.19113/sdufenbed.1575098.
JAMA
1.Şengöz N, Köroğlu H, Kırıktaş BN. Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models. J. Nat. Appl. Sci. 2025;29:124–133.
MLA
Şengöz, Nilgün, et al. “Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm With Explainable Artificial Intelligence Models”. Süleyman Demirel Üniversitesi Fen Bilimleri Enstitüsü Dergisi, vol. 29, no. 1, Apr. 2025, pp. 124-33, doi:10.19113/sdufenbed.1575098.
Vancouver
1.Nilgün Şengöz, Harun Köroğlu, Beyza Nur Kırıktaş. Detection of Rotten Fruits Using XGBoost-Based Deep Learning Algorithm with Explainable Artificial Intelligence Models. J. Nat. Appl. Sci. 2025 Apr. 1;29(1):124-33. doi:10.19113/sdufenbed.1575098

Cited By

e-ISSN :1308-6529
Linking ISSN (ISSN-L): 1300-7688

All published articles in the journal can be accessed free of charge and are open access under the Creative Commons CC BY-NC (Attribution-NonCommercial) license. All authors and other journal users are deemed to have accepted this situation. Click here to access detailed information about the CC BY-NC license.