Ceviz işleme hatları için YOLOv11 model temelli otomatik tespit ve sınıflandırma sistemleri

Öz

Tarım ürünlerinin otomatik sınıflandırılması, kalite kontrol süreçlerinin hızlandırılması ve insan hatasının azaltılması açısından kritik öneme sahiptir. Bu çalışmada konveyör bant üzerinde bulunan çoklu ceviz bileşenlerinin gerçek zamanlı olarak tespiti ve sınıflandırılması amacıyla, YOLOv11 tabanlı bir derin öğrenme yöntemi önerilmiştir. Endüstriyel bir düzenek üzerinde görüntüler alınmış ve toplamda 1194 adet kabuk, 641 adet ceviz içi ve 458 adet zar görüntüsü etiketlenmiştir. Etiketlenen bu veriler ilk önce YOLOv11n modeli ile eğitilip test edilmiş olup; en yüksek recall değeri ceviz içi sınıfında 0.963 olarak ve en düşük recall değeri zar sınıfında 0.795 olarak hesaplanmıştır. Aynı etiketli veri seti daha büyük bir model olan YOLOv11L modeli ile eğitilip test edildiğinde en yüksek recall değeri ceviz içi sınıfında 0.977 ve en düşük recall değeri zar sınıfında 0.922 olarak hesaplanmıştır. Son olarak YOLO modellerinden farklı olarak RT-DETR nesne tespit algoritması alternatif bir model olarak kullanılmış ve sonuçları YOLOv11 modelleri ile kıyaslanmıştır. Her bir modelin kendine göre üstün yanları olup; elde edilen test sonuçlarına göre yüksek fps uygulamaları için YOLOv11n modeli, orta fps ve yüksek doğruluk için YOLOv11L modeli, orta - düşük fps değerlerinde kabuk ve ceviz içi sınıflarının yüksek doğrulukta tespiti için RT-DETR modeli önerilmektedir.

Anahtar Kelimeler

• Nesne Tespiti
• Nesne Sınıflandırma
• YOLOv11
• RT-DETR
• Ceviz Tespit ve Sınıflandırma

Automatic detection and classification systems based on YOLOv11 model for walnut processing lines

Abstract

Automatic classification of agricultural products is critical for accelerating quality control processes and reducing human error. In this study, a YOLOv11-based deep learning method is proposed for real-time detection and classification of multiple walnut components on a conveyor belt. Images were captured in an industrial setup, and a total of 1194 shell, 641 kernel, and 458 membrane images were labeled. These labeled data were first trained and tested with the YOLOv11n model; the highest recall value was calculated as 0.963 in the walnut class and the lowest recall value was calculated as 0.795 in the membrane class. When the same labeled dataset was trained and tested with the larger YOLOv11L model, the highest recall value was 0.977 for the walnut kernel class, and the lowest recall value was 0.922 for the membrane class. Finally, unlike the YOLO models, the RT-DETR object detection algorithm was employed as an alternative model, and its results were compared with those of the YOLOv11 models. Each model has its own advantages; based on the test results, the YOLOv11n model is recommended for high fps applications, the YOLOv11L model for medium fps with high accuracy, and the RT-DETR model is recommended for high accuracy detection of shell and walnut kernel classes at medium - low fps values.

Keywords

• Object Detection
• Object Classification
• YOLOv11n
• RT-DETR
• Walnut Detection and Classification

Kaynakça

1. N. Dalal ve B. Triggs, Histograms of oriented gradients for human detection. 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), sayfa 886-893, San Diego, ABD, 20-25 Haziran 2005.
2. D. G. Lowe, Distinctive image features from scaleinvariant keypoints, International Journal Of Computer Vision, vol. 60, pp. 91–110, 2004.
3. N. O’Mahony, S. Campbell, A. Carvalho, S. Harapanahalli, G. V. Hernandez, L. Krpalkova, D. Riordan ve J. Walsh, Deep learning vs. traditional computer vision. Science and Information Conference, sayfa 128-144, 2019.
4. V. Pagire, M. Chavali ve A. Kale, A comprehensive review of object detection with traditional and deep learning methods. Signal Processing, 237, 110075, 2025. https://doi.org/10.1016/j.sigpro.2025.110075.
5. E. Karypidis, S. G. Mouslech, K. Skoulariki ve A. Gazis, Comparison analysis of traditional machine learning and deep learning techniques for data and image classification. arXiv preprint arXiv:2204.05983,2022. https://doi.org/10.37394/23206.2022.21.19.
6. I. R. Ward, H. Laga ve M. Bennamoun, RGB-D image-based object detection: from traditional methods to deep learning techniques. in: RGB-D Image Analysis and Processing, Springer, sayfa 169-201, 2019.
7. R. P. Haff, T. C. Pearson ve N. Toyofuku, Sorting of in-shell pistachio nuts from kernels using color imaging. Applied Engineering in Agriculture, 26 (4), 633-638, 2010. https://doi.org/10.13031/2013.32053.
8. P. Vasishth, Desai ve A. Bavarva, Image processing method for embedded optical peanut sorting. International Journal of Image, Graphics & Signal Processing, 7 (12), 2015. https://doi.org/10.5815/ijigsp.2015.12.06.

9. Z.-Q. Zhao, P. Zheng, S.-T. Xu ve X. Wu, Object detection with deep learning: A review. IEEE Transactions on Neural Networks and Learning Systems, 30 (11), 3212-3232, 2019. https://doi.org/10.1109/TNNLS.2018.2876865.
10. A. Krizhevsky, I. Sutskever ve G. E. Hinton, Imagenet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 25, 2012. https://doi.org/10.1145/3065386.
11. K. Simonyan ve A. Zisserman, Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014. https://doi.org/10.48550/arXiv.1409.1556.
12. C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke ve A. Rabinovich, Going deeper with convolutions. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, sayfa 1-9, 2015.
13. K. He, X. Zhang, S. Ren ve J. Sun, Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, sayfa 770-778, 2016.
14. M. Tan ve Q. Le, EfficientNet: Rethinking model scaling for convolutional neural networks. International Conference on Machine Learning, sayfa 6105-6114, 2019.
15. H. Aktaş, T. Kızildeniz ve Z. Ünal, Classification of pistachios with deep learning and assessing the effect of various datasets on accuracy. Journal of Food Measurement and Characterization, 16 (3), 1983-1996, 2022. https://doi.org/10.1007/s11694-022-01313-5.
16. Z. Wu, K. Luo, C. Cao, G. Liu, E. Wang ve W. Li, Fast location and classification of small targets using region segmentation and a convolutional neural network. Computers and Electronics in Agriculture, 169, 105207, 2020. https://doi.org/10.1016/j.compag.2019.105207.
17. J. Redmon, S. Divvala, R. Girshick ve A. Farhadi, You only look once: Unified, real-time object detection. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, sayfa 779-788, 2016.
18. J. Redmon ve A. Farhadi, YOLOv3: An incremental improvement. arXiv preprint arXiv:1804.02767, 2018. https://doi.org/10.48550/arXiv.1804.02767.
19. A. Bochkovskiy, C.-Y. Wang ve H.-Y. M. Liao, YOLOv4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934, 2020. https://doi.org/10.48550/arXiv.2004.10934.
20. G. Jocher, A. Stoken, J. Borovec, C. Liu, A. Hogan, L. Diaconu, J. Poznanski, L. Yu, P. Rai, R. Ferriday ve diğerleri, Ultralytics/YOLOv5: v3.0. Zenodo, 2020. https://doi.org/10.5281/zenodo.3983579.
21. C.-Y. Wang, A. Bochkovskiy ve H.-Y. M. Liao, YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, sayfa 7464-7475, 2023.
22. G. Jocher, A. Stoken, J. Borovec, C. Liu, A. Hogan, L. Diaconu, F. Ingham, J. Poznanski, J. Fang, L. Yu ve diğerleri, Ultralytics/YOLOv5: v3.1 – bug fixes and performance improvements. Zenodo, 2020. https://doi.org/10.5281/zenodo.4154370.
23. D. Guo, Z. Li, H. Shuai ve F. Zhou, Multi-Target vehicle tracking algorithm based on improved DeepSORT. Sensors, 24 (21), 7014, 2024. https://doi.org/10.3390/s24217014.
24. A. I. B. Parico ve T. Ahamed, Real time pear fruit detection and counting using YOLOv4 models and deep SORT, Sensors, 21 (14), 4803, 2021. https://doi.org/10.3390/s21144803.
25. J. Lei, W. Zheng, L. Zhang, W. Lv ve Y. Li, Detection of walnut internal quality via improved YOLOv5 and x-ray imaging. Journal of Food Process Engineering, 47 (10), e14742, 2024. https://doi.org/10.1111/jfpe.14742.
26. Y. Che, H. Bai, L. Sun, Y. Fang, X. Guo ve S. Yin, Real-Time detection of varieties and defects in moving corn seeds based on YOLO-SBWL. Agriculture, 15 (7), 685, 2025. https://doi.org/10.3390/agriculture15070685.
27. Y. Zhang, X. Wang, Y. Liu, Z. Li, H. Lan, Z. Zhang ve J. Ma, Machine Vision-Based Chinese Walnut shell--kernel recognition and separation. Applied Sciences, 13 (19), 10685, 2023. https://doi.org/10.3390/app131910685.
28. M. Wu, L. Yun, C. Xue, Z. Chen ve Y. Xia, Walnut recognition method for UAV remote sensing images. Agriculture, 14 (4), 646, 2024. https://doi.org/10.3390/agriculture14040646.
29. H. Zhang, X. Ning, H. Pu ve S. Ji, A novel approach for the non-destructive detection of shriveling degrees in walnuts using improved YOLOv5n based on X-ray images. Postharvest Biology and Technology, 214, 113007, 2024. https://doi.org/10.1016/j.postharvbio.2024.113007.
30. R. Khanam ve M. Hussain, YOLOv11: An overview of the key architectural enhancements. arXiv preprint arXiv:2410.17725, 2024. https://doi.org/10.48550/arXiv.2410.17725.
31. A. A. Lubis, A. Prasasta ve D. A. Sari, Towards efficient crowd counting and behavior analysis using YOLOv11. International Journal of Technology and Modeling, 4 (1), 35-47, 2025. https://doi.org/10.63876/ijtm.v4i1.128.
32. Y. Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y. Liu ve J. Chen, DETRs beat YOLOs on real-time object detection. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, sayfa 16965-16974, 2024.
33. L. Bustamante ve J. C. Gutiérrez, Enhancing Real-Time Detection Transformer (RT-DETR) for Handgun Detection on Nvidia Jetson. CLEI Electronic Journal, 28 (3), 5-1, 2025.