You are here: Home Аrchive of articles (from 2017 till now) 2026 №3 List of Article 04 Gabitov I. I., Safin F. R., Katkov A. A., Dmitriev А. М., Kazmiruk A. S., Kashapov A. A. A Neural Network Approach to Digital Phenotyping of Pea Sheaf Material: from Multi-Class Segmentation to Specialized Refinement Models

PDF To download article.

DOI: 10.15507/2658-4123.26363.518-540

UDK 63:633.358:004.9-045.64

 

A Neural Network Approach to Digital Phenotyping of Pea Sheaf Material: from Multi-Class Segmentation to Specialized Refinement Models

 

Ildar I. Gabitov
Dr.Sci. (Eng.), rector, Bashkir State Agrarian University (34, 50-letiya Oktyabrya St., Ufa 450001, Russian Federation), ORCID: https://orcid.org/0000-0003-4443-3126, This email address is being protected from spambots. You need JavaScript enabled to view it.

Filyus R. Safin
Dr.Sci. (Eng.), Associate Professor of the Chair of Power Supply and Automation of Technological Processes, Bashkir State Agrarian University (34, 50-letiya Oktyabrya St., Ufa 450001, Russian Federation), ORCID: https://orcid.org/0000-0003-2228-3278, Researcher ID: F-3892-2018, Scopus ID: 57217827478, SPIN-code: 2947-4312, This email address is being protected from spambots. You need JavaScript enabled to view it.

Aleksey A. Katkov
Cand.Sci. (Eng.), Associate Professor of the Department of Life Safety and Technological Equipment, Bashkir State Agrarian University (34, 50-letiya Oktyabrya St., Ufa 450001, Russian Federation), ORCID: https://orcid.org/0009-0005-2065-2705, This email address is being protected from spambots. You need JavaScript enabled to view it.

Aleksey М. Dmitriev
Cand.Sci. (Agric.), Associate Professor of the Department of Plant Growing, Plant Breeding and Biotechnology, Bashkir State Agrarian University (34, 50-letiya Oktyabrya St., Ufa 450001, Russian Federation), ORCID: https://orcid.org/0000-0003-1596-7372, This email address is being protected from spambots. You need JavaScript enabled to view it.

Alexey S. Kazmiruk
graduate student, Bashkir State Agrarian University (34, 50-letiya Oktyabrya St., Ufa 450001, Russian Federation), ORCID: https://orcid.org/0009-0003-2762-1069, This email address is being protected from spambots. You need JavaScript enabled to view it.

Artur А. Kashapov
general Director of Titanium Digital LLC (1, Rudolf Nuriev St., Ufa 450096, Russian Federation), ORCID: https://orcid.org/0009-0007-3724-8586, This email address is being protected from spambots. You need JavaScript enabled to view it.

 

Abstract
Introduction. Digital phenotyping of dried pea sheaf material is essential for accelerating breeding work in plant-growing. Automatic analysis of such samples is complicated by overlapping organs, low contrast, similar coloration, and extreme pixel imbalance between morphological classes. To solve this problem, there has been proposed a cascade neural network approach combining multi-class semantic segmentation on crops with subsequent refinement of underrepresented classes using a specialized binary model.
Aim of the Study. The study is aimed at developing a neural network approach for digital phenotyping of pea sheaf material. This approach consists in the sequential application of semantic multi-class segmentation and a specialized binary refinement model for poorly represented morphological organs.
Materials and Methods. The original dataset included 162 images of seven pea varieties, manually labeled into five classes (organs): background, stem, stipule, pod, and root. The baseline U-Net architecture with the ResNet-34 encoder was trained through using full images and 512x512 pixel crops with the use of a combined loss function. For the root, there was formed a binary root-only model with an appropriate loss function and a pos_weight parameter of 50, increasing the penalty for errors in the positive class.
Results. The quality of neural network semantic multiclass segmentation is largely determined not only by the model architecture but also by the data preparation method. Maintaining the same U-Net architecture, the transition from whole images to cropped images increased the average class recognition IoU from 42.89 to 92.01%. Meanwhile, the IoU for the minority Root class remained at 30.63%. A specialized root-only model demonstrated an increase in Root IoU for the Root class to 88.46%. This confirms that a specialized refinement model is appropriate for poorly represented and visually complex plant organs in images.
Conclusion. The obtained data show that increasing the number of epochs alone does not eliminate the impact of extreme class imbalance across pixels. Separating the general and specialized segmentation tasks is the most effective. The proposed cascade neural network approach can be used as the basis for a digital phenotyping software suite and adapted to other crops characterized by small, low-contrast, and sparsely represented organs.

Keywords: pea, morphological features, computer vision, digital phenotyping, semantic segmentation, neural network

Conflict of interest: The authors declare that there is no conflict of interest.

For citation: Gabitov I.I., Safin F.R., Katkov A.A., Dmitriev А.М., Kazmiruk A.S., Kashapov A.A. A Neural Network Approach to Digital Phenotyping of Pea Sheaf Material: from Multi-Class Segmentation to Specialized Refinement Models. Engineering Technologies and Systems. 2026;36(3):518–540. https://doi.org/10.15507/2658-4123.26363.518-540

Authors contribution:
I. I. Gabitov – oversight and leadership responsibility for the research activity planning and execution, including mentorship external to the core team; application of statistical, mathematical, computational, or other formal techniques to analyses or synthesize study data; creation of the published work.
F. R. Safin – application of statistical, mathematical, computational, or other formal techniques to analyses or synthesize study data; conducting a research and investigation process, specifically performing the experiments, or data collection; verification, whether as a part of the activity or separate, of the overall replication / reproducibility of results / experiments and other research outputs.
A. A. Katkov – preparation, creation of the published work; revision of the initial text; preparation, creation and of the published work by those from the original research group, specifically critical review, commentary or revision – including pre- or post-publication stages.
A. M. Dmitriev – verification, whether as a part of the activity or separate, of the overall replication / reproducibility of results / experiments and other research outputs.
A. S. Kazmiruk – conducting a research and investigation process, specifically performing the experiments, or data collection.
A. A. Kashapov – management activities to annotate (produce metadata), scrub data and maintain research data (including software code, where it is necessary for interpreting the data itself) for initial use and later re-use.

All authors have read and approved the final manuscript.

Submitted 02.07.2026;
revised 16.07.2026;
accepted 20.07.2026

 

REFERENCES

  1. Li L., Zhang Q., Huang D. A Review of Imaging Techniques for Plant Phenotyping. Sensors. 2014;14(11):20078–20111. https://doi.org/10.3390/s141120078
  2. Zhao C., Zhang Y., Du J., Guo X., Wen W., Gu S., et al. Crop Phenomics: Current Status and Perspectives. Frontiers in Plant Science. 2019;(10):714. https://doi.org/10.3389/fpls.2019.00714
  3. Shashko A.Yu., Bondarenko V.Yu., Chernysh M.A., Przhevalskaya D.A., Usnich S.L., Pshibytko N.L., et al. Modern Phenomic Platforms and their Application in Plant Biology and Agriculture. Journal of the Belarusian State University. Biology. 2020;(2):15–25. (In Russ., abstract in Eng.) https://doi.org/10.33581/2521-1722-2020-2-15-25
  4. Kuznetsov I.Yu., Dmitriev A.M., Avsakhov F.F., Alimgafarov R.R., Rodionov N.Yu. Adaptive Potential of Collection Samples of Field Pea: Assessmentf Plasticity and Stability in the Conditions of the Republic of Bashkortostan. Legumes and Groat Crops. 2026;1(57):17–26. (In Russ., abstract in Eng.) https://doi.org/10.24412/2309-348X-2026-1-17-26
  5. Daba S.D., Morris C.F. Pea Proteins: Variation, Composition, Genetics, and Functional Properties. Cereal Chemistry. 2022;99(1):8–20. https://doi.org/10.1002/cche.10439
  6. Gabitov I.I., Safin F.R., Katkov A.A., Shamukaev S.B., Kazmiruk A.C. Development of Computer Vision Algorithms for Digital Phenotyping of Peas. Vestnik of the Bashkir State Agrarian University. 2026;(1):109–113. (In Russ., abstract in Eng.) https://doi.org/10.31563/1684-7628-2026-77-1-109-113
  7. Furbank R.T., Tester M. Phenomics – Technologies to Relieve the Phenotyping Bottleneck. Trends in Plant Science. 2011;16(12):635–644. https://doi.org/10.1016/j.tplants.2011.09.005
  8. Fiorani F., Schurr U. Future Scenarios for Plant Phenotyping. Annual Review of Plant Biology. 2013;(64):267–291. https://doi.org/10.1146/annurev-arplant-050312-120137
  9. Luo Z., Yang W., Yuan Y., Gou R., Li X. Semantic Segmentation of Agricultural Images: A Survey. Information Processing in Agriculture. 2024;11(2):172–186. https://doi.org/10.1016/j.inpa.2023.02.001
  10. Lei L., Yang Q., Yang L., Shen T., Wang R., Fu C. Deep Learning Implementation of Image Segmentation in Agricultural Applications: A Comprehensive Review. Artificial Intelligence Review. 2024;(57):149. https://doi.org/10.1007/s10462-024-10775-6
  11. Mudarisov S.G., Miftakhov I.R., Farkhutdinov I.M., Shafeeva E.I. Development of a Low-Power Onboard System for Spot Detection of Wheat Phytopathologies Using a Modified YOLO Architecture. Engineering Technologies and Systems. 2026;36(1):114–139. (In Russ., abstract in Eng.) https://doi.org/10.15507/2658-4123.036.202601.114-139
  12. Arinichev I.V., Polyanskikh S.V., Arinicheva I.V. Semantic Segmentation of Rusts and Spots of Wheat. Computer Optics. 2023;47(1):118–125. (In Russ., abstract in Eng.) https://doi.org/10.18287/2412-6179-CO-1130
  13. Liu F., Yang R., Chen R., Guindo M.L., He Y., Zhou J., et al. Digital Techniques and Trends for Seed Phenotyping Using Optical Sensors. Journal of Advanced Research. 2024;(63):1–16. https://doi.org/10.1016/j.jare.2023.11.010
  14. Warman C., Sullivan C.M., Preece J., Buchanan M.E., Vejlupkova Z., Jaiswal P., et al. A Cost Effective Maize Ear Phenotyping Platform Enables Rapid Categorization and Quantification of Kernels. The Plant Journal. 2021;106(2):566–579. https://doi.org/10.1111/tpj.15166
  15. Minervini M., Fischbach A., Scharr H., Tsaftaris S.A. Finely-Grained Annotated Datasets for Image-Based Plant Phenotyping. Pattern Recognition Letters. 2016;(81):80–89. https://doi.org/10.1016/j.patrec.2015.10.013
  16. Scharr H., Minervini M., French A.P., Klukas C., Kramer D.M., Liu X., et al. Leaf Segmentation in Plant Phenotyping: A Collation Study. Machine Vision and Applications. 2016;(27):585–606. https://doi.org/10.1007/s00138-015-0737-3
  17. Ubbens J.R., Stavness I. Deep Plant Phenomics: A Deep Learning Platform for Complex Plant Phenotyping Tasks. Frontiers in Plant Science. 2017;(8):1190. https://doi.org/10.3389/fpls.2017.01190
  18. Pound M.P., Burgess A.J., Wilson M.H., Atkinson J.A., Griffiths M., Jackson A.S. Deep Machine Learning Provides State-of-the-Art Performance in Image-Based Plant Phenotyping. GigaScience. 2016. https://doi.org/10.1101/053033
  19. Bhagat S., Kokare M., Haswani V., Hambarde P., Kamble R. Eff-UNet++: A Novel Architecture for Plant Leaf Segmentation and Counting. Ecological Informatics. 2022;(68):101583. https://doi.org/10.1016/j.ecoinf.2022.101583
  20. Rakutko S.A., Rakutko E.N., Su J. Estimating Chlorophyll Content by Optical Density of Plant Leaves Using Machine Learning. Engineering Technologies and Systems. 2025;35(4):678–699. (In Russ., abstract in Eng.) https://doi.org/10.15507/2658-4123.035.202504.678-699
  21. Belousov I.S., Rogachev A.F. Creation a Neural Network Service for Identifying Problem Areas of Agricultural Fields. Proc. of the Lower Volga Agro-University Comp. 2021;3(63):439–448. (In Russ., abstract in Eng.) https://doi.org/10.32786/2071-9485-2021-03-45
  22. Molin A.E., Blekanov I.S., Mitrofanov E.P., Mitrofanova O.A. Synthetic Data Generation Methods for Training Neural Networksin the Task of Segmenting the Level of Crop Nitrogen Status in Imagesof Unmanned Aerial Vehicles in an Agricultural Field. Vestnik of Saint Petersburg University. Applied Mathematics. Computer Science. Control Processes. 2024;20(1):20–33. (In Russ., abstract in Eng.) https://doi.org/10.21638/11701/spbu10.2024.103
  23. Yasrab R., Atkinson J.A., Wells D.M., French A.P., Pridmore T.P., Pound M.P. RootNav 2.0: Deep Learning for Automatic Navigation of Complex Plant Root Architectures. GigaScience. 2019;8(11):123. https://doi.org/10.1093/gigascience/giz123
  24. Smith A.G., Han E., Petersen J., Olsen N.A.F., Giese C., Athmann M., et al. RootPainter: Deep Learning Segmentation of Biological Images With Corrective Annotation. New Phytologist. 2022;236(2):774–791. https://doi.org/10.1111/nph.18387
  25. Atkinson J.A., Pound M.P., Bennett M.J., Wells D.M. Uncovering the Hidden Half of Plants Using New Advances in Root Phenotyping. Current Opinion in Biotechnology. 2018;(55):1–8. https://doi.org/10.1016/j.copbio.2018.06.002
  26. Kuijken R.C.P., Eeuwijk F.A., Marcelis L.F.M., Bouwmeester H.J. Root Phenotyping: from Component Trait in the Lab to Breeding. Journal of Experimental Botany. 2015;66(18):5389–5401. https://doi.org/10.1093/jxb/erv239
  27. Carranza-Rojas J., Goëau H., Bonnet P., Mata-Montero E., Joly A. Going Deeper in the Automated Identification of Herbarium Specimens. BMC Evolutionary Biology. 2017;(17):181. https://doi.org/10.1186/s12862-017-1014-z
  28. Ronneberger O., Fischer P., Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. MICCAI 2015: Lecture Notes in Computer Science. Springer; Cham. 2015. pp. 234–241. https://doi.org/10.1007/978-3-319-24574-4_28
  29. He K., Zhang X., Ren S., Sun J. Deep Residual Learning for Image Recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE. 2016. pp. 770–778. https://doi.org/10.1109/CVPR.2016.90
  30. Buda M., Maki A., Mazurowski M.A. A Systematic Study of the Class Imbalance Problem in Convolutional Neural Networks. Neural Networks. 2018;(106):249–259. https://doi.org/10.1016/j.neunet.2018.07.011
  31. Sudre C.H., Li W., Vercauteren T., Ourselin S., Cardoso M.J. Generalised Dice overlap as a Deep Learning Loss Function for Highly Unbalanced Segmentations. In: Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. Lecture Notes in Computer Science. 2017. pp. 240–248. https://doi.org/10.1007/978-3-319-67558-9_28

 

Licensed under a Creative Commons
This work is licensed under a Creative Commons Attribution 4.0 License.

Details
Hits: 3