Logo PTI Logo FedCSIS

Position Papers of the 21st Conference on Computer Science and Intelligence Systems

Annals of Computer Science and Information Systems, Volume 48

Improved Estimation of Soil-to-Sugar Beet Ratio Using Deep Learning and Domain Adaptation for Precision Agriculture

, , , ,

DOI: http://dx.doi.org/10.15439/2026F2778

Citation: Elham Alfuqara, , , ,

Full text

Abstract. Accurately determining the amount of soil adhering to sugar beets is critical for both sugar companies and farmers. Ideally, net weight would be obtained by weighing the beets before and after cleaning for each truckload, but logistical challenges make this impractical. As an alternative, companies and farmers rely on trained human estimators who visually assess soil-to-beet ratio during unloading, considering factors such as soil type (dry or wet) and soil layer thickness. However, due to the subjective nature of these assessments, Gage R\&R and accuracy studies have been conducted, highlighting variability among estimators and the need for a consistent, automated solution. This paper presents a processing pipeline to improve soil-to-beet ratio estimation reliability, starting with sugar beet segmentation using Detectron2, followed by soil masking via custom image processing techniques that exploit intensity and texture information. Features extracted from the segmented soil regions are used to train a Semi-Supervised Domain-Adversarial Neural Network (SS-DANN), designed to address domain shifts between the ground truth and field datasets. The ground truth dataset consists of lab-captured images of individual beets with precisely measured soil weights, while the field dataset includes infrared images acquired during real loading operations. After training, the proposed method achieved an average deviation of 0.39\% from the ground truth, compared to 1.91\% average deviation by human estimators, demonstrating a significant improvement in accuracy and consistency.

References

  1. G. D. Vermeulen, J. J. Klooster, M. C. Sprong, and B. R. Verwijs, "Soil Tare and Relative Soil Adherence after Uprooting Sugar Beet by a Share Lifter, a Driven Rotary-shoe Lifter and a Grab Lifter," 2003, https://doi.org/10.1016/S1537-5110(03)00117-X.
  2. P. Panagos, P. Borrelli, and J. Poesen, "Soil loss due to crop harvesting in the European Union: A first estimation of an underrated geomorphic process,Science of the Total Environment, vol. 664, pp. 487–498, 2019,https://doi.org/10.1016/j.scitotenv.2019.02.009.
  3. M. Zengin, R. Uyanöz, and Ü. Çetin, “A Study on The Soil Tare of Sugar Beet in Konya”, Selcuk J Agr Food Sci, vol. 17, no. 31, pp. 53–55, June 2003, [Online]. Available: https://izlik.org/JA26LW26UB
  4. CIBE and CEFS, The EU Beet and Sugar Sector: A Model of Environmental Sustainability. Brussels, Belgium, Apr. 2010.
  5. M. Asadi, Beet-Sugar Handbook. Hoboken, NJ, USA: Wiley-Interscience, 2006, https://doi.org/10.1002/0471790990.
  6. T. Bakker, H. Wouters, K. van Asselt, J. Bontsema, L. Tang, J. Müller, and G. van Straten, "A vision based row detection system for sugar beet," Computers and Electronics in Agriculture, vol. 60, no. 1, pp. 87–95, 2008, https://doi.org/10.1016/j.compag.2007.07.006.
  7. A. Milioto, P. Lottes, and C. Stachniss, "Real-Time Blob-Wise Sugar Beets vs Weeds Classification for Monitoring Fields Using Convolutional Neural Networks," ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, vol. IV-2/W3, pp. 41–48, 2017, https://doi.org/10.5194/isprs-annals-IV-2-W3-41-2017.
  8. A. Jafari, S. S. Mohtasebi, H. E. Jahromi, and M. Omid, "Weed Detection in Sugar Beet Fields Using Machine Vision, International Journal of Agriculture and Biology,vol. 8, no. 5, pp. 602–605, 2006.
  9. J. Gao, A. P. French, M. P. Pound, Y. He, T. P. Pridmore, and M. J. Pieters, "Deep convolutional neural networks for image-based Convolvulus sepium detection in sugar beet fields," Plant Methods, vol. 16, Art. no. 29, 2020, https://doi.org/10.1186/s13007-020-00570-z.
  10. K. Adem, M. M. Ozguven, and Z. Altas, "A sugar beet leaf disease classification method based on image processing and deep learning," Multimedia Tools and Applications, vol. 82, pp. 12577–12594, 2023, https://doi.org/10.1007/s11042-022-13925-6.
  11. L. Pan, Q. Zhu, R. Lu, and J. M. McGrath, "Determination of sucrose content in sugar beet by portable visible and near-infrared spectroscopy,", Food Chemistry, vol. 167, pp. 264–271, 2015, https://doi.org/10.1016/j.foodchem.2014.06.117.
  12. Z. Guo, H. H. Goh, X. Li, M. Zhang, and Y. Li, "WeedNet-R: A sugar beet field weed detection algorithm based on enhanced RetinaNet and context semantic fusion,", Frontiers in Plant Science, vol. 14, Art. no. 1226329, 2023, https://doi.org/10.3389/fpls.2023.1226329.
  13. A. Nasiri, S. Taheri-Garavand, Y. Zhang, and Y.-D. Zhang, "Deep learning-based precision agriculture through weed recognition in sugar beet fields,", Sustainable Computing: Informatics and Systems, vol. 35, Art. no. 100759, 2022, https://doi.org/10.1016/j.suscom.2022.100759.
  14. A. Nasirahmadi, U. Wilczek, and O. Hensel,"Sugar beet damage detection during harvesting using different convolutional neural network models,", Agriculture,vol. 11, no. 11, Art. no. 1111, 2021, https://doi.org/10.3390/agriculture11111111.
  15. G. Croonen, A. Trondl, J. Simon, and D. Steininger, "SemanticSugarBeets: A Multi-Task Framework and Dataset for Inspecting Harvest and Storage Characteristics of Sugar Beets,", arXiv preprint https://arxiv.org/abs/2504.16684, 2025, https://doi.org/10.48550/arXiv.2504.16684.
  16. JAI, FS-3200T-10GE-NNC 3-CMOS area scan camera, JAI, [Online]. Available: https://www.jai.com/products/fs-3200t-10ge-nnc/ [Accessed: May 16, 2025].
  17. W. O. S. Soares, R. S. Peruchi, R. A. V. Silva, and P. Rotella Junior,"Gage R&R studies in measurement system analysis: A systematic literature review,", Quality Engineering, vol. 34, no. 3, pp. 382–403, 2022, https://doi.org/10.1080/08982112.2022.2069505.
  18. D. C. Montgomery, Introduction to Statistical Quality Control, 7th ed., Hoboken, NJ: John Wiley & Sons, 2012.
  19. Automotive Industry Action Group (AIAG), Measurement Systems Analysis (MSA) Reference Manual, 4th ed., Southfield, MI: AIAG, 2010.
  20. R. Girshick, I. Radosavovic, G. Gkioxari, P. Dollár, and K. He, “Detectron,” Facebook AI Research, 2018. [Online]. Available:https: //github.com/facebookresearch/detectron
  21. T. Ojala, M. Pietikainen and T. Maenpaa, “Multiresolution grayscale and rotation invariant texture classification with local binary patterns,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 7, pp. 971–987, 2002, https://doi.org/10.1109/TPAMI.2002.1017623.
  22. N. Otsu, “A Threshold Selection Method from Gray-Level Histograms,” IEEE Transactions on Systems, Man, and Cybernetics, vol. 9, no. 1, pp. 62–66, 1979, https://doi.org/10.1109/TSMC.1979.4310076.
  23. P. Soille, Morphological Image Analysis: Principles and Applications, 2nd ed. Springer, https://doi.org/10.1007/978-3-662-05088-0.
  24. R. M. Haralick, K. Shanmugam, and I. Dinstein, “Textural features for image classification,” IEEE Transactions on Systems, Man, and Cybernetics, vol. SMC-3, no. 6, pp. 610–621, 1973, https://doi.org/10.1109/TSMC.1973.4309314.
  25. B. S. Manjunath and W. Y. Ma, “Texture features for browsing and retrieval of image data,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 18, no. 8, pp. 837–842, 1996, https://doi.org/10.1109/34.531803.
  26. M. Wang and W. Deng, "Deep visual domain adaptation: A survey,", Neurocomputing, vol. 312, pp. 135–153, Oct. 2018, https://doi.org/10.1016/j.neucom.2018.05.083.
  27. O. Kilim, A. Olar, T. Joó, T. Palicz, P. Pollner, and I. Csabai, “Physical imaging parameter variation drives domain shift,” Scientific Reports, vol. 12, Art. no. 21302, 2022, https://doi.org/10.1038/s41598-022-239904.
  28. Y. Ganin, E. Ustinova, H. Ajakan, P. Germain, H. Larochelle, F. Laviolette, M. Marchand, and V. Lempitsky, “Domain-Adversarial Training of Neural Networks,” arXiv preprint arXiv:1505.07818, https://doi.org/10.48550/arXiv.1505.07818.
  29. T. Hastie, R. Tibshirani, and J. Friedman, "The Elements of Statistical Learning: Data Mining, Inference, and Prediction,", 2nd ed. New York, NY, USA: Springer, 2009, https://doi.org/10.1007/978-0-387-848587.
  30. S. H. Khan, M. Hayat, M. Bennamoun, F. A. Sohel, and R. Togneri, "Cost-Sensitive Learning of Deep Feature Representations From Imbalanced Data,", IEEE Transactions on Neural Networks and Learning Systems, vol. 29, no. 8, pp. 3573–3587, Aug. 2018, https://doi.org/10.1109/TNNLS.2017.2732482.
  31. T. Lin et al., "Focal Loss for Dense Object Detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 42, no. 2, pp. 318-327, 2020, https://doi.org/10.48550/arXiv.1708.02002.
  32. B. Sun and K. Saenko, “Deep CORAL: Correlation Alignment for Deep Domain Adaptation,” in Proc. ECCV Workshop, 2016, pp. 443–450, https://doi.org/10.48550/arXiv.1607.01719