DEVELOPMENT OF AN AUTOMATED METHOD FOR FORECASTING STRAWBERRY YIELD USING UAVS AND MACHINE LEARNING ALGORITHMS
DOI 10.31588/2073-0462-2026-21-1-69-78 УДК / UDC 631.171.004.3:004.932 EDN: LIDPSI
Keywords:
convolutional neural networks, transfer learning, image processing, monitoring, yield forecasting, precision farming.Abstract
This article presents a method for remote monitoring and forecasting of strawberry (Fragaria × ananassa) yields based on unmanned aerial vehicles (UAVs), computer vision and deep learning technologies. Using morphological features of the crop, UAVs ensure high detection accuracy. The study was conducted using YOLO12 Extra Large convolutional neural network, trained using transfer learning on a specialized dataset of strawberry flower images (class “Flower”). The dataset was created by manual labeling in RectLabel and expanded to 1700 images (18 496 exemplars) using complex data augmentation. Initial data were collected using a DJI Mavic 3M UAV (RGB camera) at an altitude of 30-60 m. Image processing included the construction of a high-precision orthomosaic (resolution 0.570 mm/pixel) in Agisoft Metashape Pro. Orthomosaic was automatically divided into 200×200 pixel fragments for detection. Training YOLO12 model (hyperparameters: batch size = 8, learning rate = 0.001, epochs = 1500) achieved high metrics on the test set: precision 0.903, recall 0.812, Fmeasure 0.852, mAP50 0.888. Based on the developed method, specialized software with a graphical interface was created. It implements the full processing cycle: orthomosaic tiling, flower detection, merging detection results into a single orthomosaic, spatial analysis (construction of density heat maps, calculation of Moran’s index for clustering assessment, K-means cluster analysis, visualization of Voronoy diagrams), and yield forecasting using a formula that takes into account the number of recognized flowers, varietal conversion factors and average berry weight. The method enables automated flower counting over large areas, identification of zones of uneven productivity and yield forecasting in the early stages of vegetation, significantly reducing labor costs and providing a basis for making agricultural decisions.
References
1. Динамика производства ягод земляники садовой по странам мира / В. Н. Ожерельев, М. В. Ожерельева, А. М. Гринь и др. // Вестник Брянской государственной сельскохозяйственной академии. 2019. №4(74). С. 60–66.
2. Марченко Л.А. Продуктивность земляники садовой и селекционные возможности ее повышения // Сибирский вестник сельскохозяйственной науки. 2021. №51(3). С. 65–74. doi: 10.26898/0370-8799-2021-3-7
3. Кутырёв А.И., Филиппов Р.А. Применение сверточной нейронной сети для мониторинга состояния земляники садовой // Аграрная наука Евро-Северо-Востока. 2023. Т.24. №4. С. 685–696. doi: 10.30766/2072-9081.2023.24.4.685-696
4. Разработка системы автоматизированного управления агротехнологиями в садоводстве / Д.О. Хорт, А.И. Кутырев, И.Г. Смирнов и др. // Сельскохозяйственные машины и технологии. 2021. Т.15. №2. С. 61–68. doi: 10.22314/2073-7599-2021-15-2-61-68
5. Кутырёв А.И., Смирнов И.Г., Андриянов Н.А. Сравнительный анализ моделей нейронных сетей для распознавания плодов яблони на кроне дерева // Садоводство и виноградарство. 2023. №5. С.56–63. doi: 10.31676/0235-2591-2023-5-56-63
6. VM-YOLO: YOLO with VMamba for Strawberry Flowers Detection / Y. Wang, X. Lin, Z. Xiang, et al. // Plants. 2025. 14(3):468. doi: 10.3390/plants14030468
7. Strawberry Detection and Ripeness Classification Using YOLOv8+ Model and Image Processing Method / C. Wang, H. Wang, Q. Han, et al. // Agriculture 2024. 14(5):751. doi: 10.3390/agriculture14050751
8. Estimation of Strawberry Canopy Volume in Unmanned Aerial Vehicle RGB Imagery Using an Object Detection-Based Convolutional Neural Network / M-S. Gang, T. Sutthanonkul, W.S. Lee, et al. // Sensors 2024. 24 (21):6920. doi: 10.3390/s24216920
9. Strawberry Yield Prediction Based on a Deep Neural Network Using High-Resolution Aerial Orthoimages / Y. Chen, W.S. Lee, H. Gan, et al. // Remote Sensing. 2019. 11(13):1584. doi: 10.3390/rs11131584
10. Improving Strawberry Yield Prediction by Integrating Ground-Based Canopy Images in Modeling Approaches / A. Abd-Elrahman, F. Wu, S. Agehara, et al. // ISPRS International Journal of Geo-Information. 2021. 10 (4):239. doi: 10.3390/ijgi10040239
11. Deep learning for strawberry canopy delineation and biomass prediction from high-resolution images / C. Zheng, A. Abd-Elrahman, V.M. Whitaker, et al. // Plant Phenomics. 2022. 9850486. doi: 10.34133/2022/9850486
12. Yue Y., Xu S., Wu H.A. Strawberry Ripeness Detection Method Based on Improved YOLOv8 // Applied Sciences. 2025. 15(11):6324. doi: 10.3390/app15116324
13. YOLOv5-ASFF: A Multistage Strawberry Detection Algorithm Based on Improved YOLOv5 / Y. Li, J. Xue, M. Zhang, et al. // Agronomy. 2023. 13(7):1901. doi: 10.3390/agronomy13071901
14. Lightweight Improved YOLOv5s-CGhostnet for Detection of Strawberry Maturity Levels and Counting / N. Tamrakar, S. Karki, MY. Kang, et al. // AgriEngineering. 2024. 6(2):962-978. doi: 10.3390/agriengineering6020055
15. Research on the Strawberry Recognition Algorithm Based on Deep Learning / Y. Zhang, L. Zhang, H. Yu, et al. // Applied Sciences. 2023. 13(20):11298. doi: 10.3390/app132011298
16. Prediction of strawberry yield based on receptacle detection and Bayesian inference / S. Yoon, J.S. Jo, S.B. Kim, et al. // Heliyon. 2023. Vol. 9. No. 3. P. e14546. doi: 10.1016/j.heliyon.2023.e14546
17. Diwan T., Anirudh G., Tembhurne J.V. Object detection using YOLO: Challenges, architectural successors, datasets andapplications // Multimed. Tools Appl. 2023. No. 82. P. 9243-9275. doi: 10.1007/s11042-022-13644-y
18. D-YOLO: A Lightweight Model for Strawberry Health Detection / E. Wu, R. Ma, D. Dong, et al. // Agriculture. 2025. 15(6):570. doi: 10.3390/agriculture15060570
19. A deep-level region-based visual representation architecture for detecting strawberry flowers in an outdoor field / P.Lin, W.S.Lee, Y.M.Chen, et al. // Precision Agric. 2020. 21. P. 387–402. doi: 10.1007/s11119-019-09673-7
20. Maxwell A.E., Warner T.A., Guillén L.A. Accuracy Assessment in Convolutional Neural Network-Based Deep Learning Remote Sensing Studies - Part 1: Literature Review // Remote Sensing. 2021. 13(13):2450. doi: 10.3390/rs1313245
1. Ozherelev VN, Ozhereleva MV, Grin AM. [Dynamics of garden strawberry production in the countries of the world]. Vestnik Bryanskoy gosudarstvennoy selskokhozyaystvennoy akademii. 2019; 4(74). 60-66 p.
2. Marchenko L.A. [Productivity of garden strawberries and breeding potential for their improvement]. Sibirskiy vestnik selskokhozyaystvennoy nauki. 2021; 51(3). 65-74 p. doi: 10.26898/0370-8799-2021-3-7
3. Kutyrev AI, Filippov RA. [Application of convolutional neural network for monitoring the condition of garden strawberries]. Agrarnaya nauka Evro-Severo-Vostoka. 2023; Vol.24. 4. 685-696 p. doi: 10.30766/2072-9081.2023.24.4.685-696
4. Khort DO, Kutyrev AI, Smirnov IG. [Development of an automated management system for agricultural technologies in horticulture]. Selskokhozyaystvennye mashiny i tekhnologii. 2021; Vol.15. 2. 61-68 p. doi: 10.22314/2073-7599-2021-15-2-61-68
5. Kutyrev AI, Smirnov IG, Andriyanov NA. [Comparative analysis of neural network models for recognizing apple fruits on a tree crown]. Sadovodstvo i vinogradarstvo. 2023; 5. 56-63 p. doi: 10.31676/0235-2591-2023-5-56-63
6. Wang Y, Lin X, Xiang Z, VM-YOLO: YOLO with VMamba for strawberry flowers detection. Plants. 2025; 14(3). 468 p. doi: 10.3390/plants14030468
7. Wang C, Wang H, Han Q. strawberry detection and ripeness classification using YOLOv8+ model and image processing method. Agriculture. 2024; 14(5). 751 p. doi: 10.3390/agriculture14050751
8. Gang M-S, Sutthanonkul T, Lee WS. Estimation of strawberry canopy volume in unmanned aerial vehicle RGB imagery using an object detection-based convolutional neural network. Sensors. 2024; 24(21). 6920 p. doi: 10.3390/s24216920
9. Chen Y, Lee WS, Gan H. Strawberry yield prediction based on a deep neural network using high-resolution aerial orthoimages. Remote Sensing. 2019; 11(13). 1584 p. doi: 10.3390/rs11131584
10. Abd-Elrahman A, Wu F, Agehara S. Improving strawberry yield prediction by integrating ground-based canopy images in modeling approaches. ISPRS International Journal of Geo-Information. 2021; 10(4). 239 p. doi: 10.3390/ijgi10040239
11. Zheng C, Abd-Elrahman A, Whitaker VM. Deep learning for strawberry canopy delineation and biomass prediction from high-resolution images. Plant Phenomics. 2022; 9850486. doi: 10.34133/2022/9850486
12. Yue Y, Xu S, Wu HA. Strawberry ripeness detection method based on improved YOLOv8. Applied Sciences. 2025; 15(11). 6324 p. doi: 10.3390/app15116324
13. Li Y, Xue J, Zhang M. YOLOv5-ASFF: a multistage strawberry detection algorithm based on improved YOLOv5. Agronomy. 2023; 13(7). 1901 p. doi: 10.3390/agronomy13071901
14. Tamrakar N, Karki S, Kang MY. Lightweight improved YOLOv5s-CGhostnet for detection of strawberry maturity levels and counting. AgriEngineering. 2024; 6(2). 962-978 p. doi: 10.3390/agriengineering6020055
15. Zhang Y, Zhang L, Yu H. Research on the strawberry recognition algorithm based on deep learning. Applied Sciences. 2023; 13(20). 11298 p. doi: 10.3390/app132011298
16. Yoon S, Jo JS, Kim SB. Prediction of strawberry yield based on receptacle detection and Bayesian inference. Heliyon. 2023; Vol.9. No.3. e14546 p. doi: 10.1016/j.heliyon.2023.e14546
17. Diwan T, Anirudh G, Tembhurne JV. Object detection using YOLO: challenges, architectural successors, datasets and applications. Multimed. Tools Appl. 2023; No.82. 9243-9275 p. doi: 10.1007/s11042-022-13644-y
18. Wu E, Ma R, Dong D. D-YOLO: a lightweight model for strawberry health detection. Agriculture. 2025; 15(6). 570 p. doi: 10.3390/agriculture15060570
19. Lin P, Lee WS, Chen YM. A deep-level region-based visual representation architecture for detecting strawberry flowers in an outdoor field. Precision Agric. 2020; 21. 387-402 p. doi: 10.1007/s11119-019-09673-7
20. Maxwell AE, Warner TA, Guillen LA. Accuracy assessment in convolutional neural network-based deep learning remote sensing studies - Part 1: literature review. Remote Sensing. 2021; 13(13). 2450 p. doi: 10.3390/rs1313245
Downloads
Published
How to Cite
Issue
Section
Categories
License
Copyright (c) 2026 Vestnik of Kazan state agrarian university
This work is licensed under a Creative Commons Attribution 4.0 International License.
Метаданные научных статей, включая название, аннотацию, сведения об авторах, список литературы, идентификаторы и иные элементы библиографического описания для дальнейшего неограниченного использования, передаются в редакцию с правом их неограниченного использования на условиях открытой лицензии Creative Commons Attribution (CC BY).
