Application of computer vision and deep learning algorithms for tree inventory in intensive orchards
https://doi.org/10.26898/0370-8799-2026-2-10
Abstract
A method for monitoring plant extinction (dead trees with signs of crown and trunk drying) in intensive orchards is presented based on aerial photography of trees from a DJI Mavic 3 Multispectral unmanned aerial vehicle in row spacing (shooting height 1.5–2.0 m) and computer vision and deep learning algorithms. An original dataset of 5,366 images of apple orchards was compiled, labeled into three classes: dead trees (class "Dead"), candidates for plant extinction (class "Candidate"), and young plantings (class "Newly"). Data augmentation (synthetic generation of new images based on the existing ones) included modeling of the changes in illumination (±15%), color tone (–20° in hue), Gaussian noise (0.5%), and blur (σ = 0.5 pixels), which increased the model’s robustness to work in garden planting conditions. Five versions of the YOLOv12 architecture were compared; the YOLOv12l model was recognized as optimal (mAP@50(B) = 0.810; Precision = 0.820; Recall = 0.831), providing a balance between accuracy and performance (9.58 ms/frame). Specialized software with a graphical interface was developed that implements video stream post-processing, object tracking, and visualization of results. The use of the ByteTrack algorithm (with camera motion compensation) allowed reducing the number of ID switches of the recognized tree classes to 2.3 per 1000 frames and increasing the recognition accuracy to 92.4%, which is 41.0% higher than the recognition results without tracking. The program supports customization of the frame analysis zone depending on garden parameters (row spacing, shooting angle) and is compatible with geoinformation platforms, allowing the creation of digital maps of crop losses referenced to GNSS-RTK coordinates. The practical significance of the obtained results lies in the reduction of labor costs for inventory, and the increase in the accuracy and speed of monitoring the condition of the plantings.
Keywords
Supporting agencies: The research was carried out under the support of the Ministry of Science and Higher Education of the Russian Federation within the state assignment of the Federal Scientific Agroengineering Center VIM (theme No. FGUN-2022-0011).
About the Author
A. I. Kutyrev
Federal Scientific Agroengineering Center VIM
Russian Federation
Russian Federation
Alexey I. Kutyrev, Laboratory Head, Lead Researcher
5, 1st Institute passage, Moscow, 109428
References
1. Laužikė K., Uselis N., Kviklys D., Samuolienė G. Orchard Planting Density and Tree Development Stage Affects Physiological Processes of Apple (Malus domestica Borkh.) Tree. Agronomy, 2020, vol. 10 (12), p. 1912. DOI: 10.3390/agronomy10121912.
2. Starkus A., Morkūnaitė-Haimi Š., Gurskas T., Misiukevičius E., Stanys V., Frercks B. The Biological and Genetic Mechanisms of Fruit Drop in Apple Tree (Malus × domestica Borkh.). Horticulturae, 2024, vol. 10 (9), p. 987. DOI: 10.3390/horticulturae10090987.
3. Sekharamantry P.K., Melgani F., Malacarne J., Ricci R., de Almeida Silva R., Marcato Junior J. A Seamless Deep Learning Approach for Apple Detection, Depth Estimation, and Tracking Using YOLO Models Enhanced by Multi-Head Attention Mechanism. Computers, 2024, vol. 13 (3), p. 83. DOI: 10.3390/computers13030083.
4. Panigrahi S.S., Singh K.D., Balasubramanian P., Wang H., Natarajan M., Ravichandran P. UAVBased LiDAR and Multispectral Imaging for Estimating Dry Bean Plant Height, Lodging and Seed Yield. Sensors, 2025, vol. 25 (11), p. 3535. DOI: 10.3390/s25113535.
5. Jemaa H., Bouachir W., Leblon B., LaRocque A., Haddadi A., Bouguila N. UAV-Based Computer Vision System for Orchard Apple Tree Detection and Health Assessment. Remote Sens, 2023, vol. 15 (14), p. 3558. DOI: 10.3390/rs15143558.
6. Hobart M., Pflanz M., Tsoulias N., Weltzien C., Kopetzky M., Schirrmann M. Fruit Detection and Yield Mass Estimation from a UAV Based RGB Dense Cloud for an Apple Orchard. Drones, 2025, vol. 9 (1), p. 60. DOI: 10.3390/drones9010060.
7. Yang R., Yuan D., Zhao M., Zhao Z., Zhang L., Fan Y., Liang G., Zhou Y. Camellia oleifera Tree Detection and Counting Based on UAV RGB Image and YOLOv8. Agriculture, 2024, vol. 14 (10), p. 1789. DOI: 10.3390/agriculture14101789.
8. Han P., Ma C., Chen J., Chen L., Bu S., Xu S., Zhao Y., Zhang C., Hagino T. Fast Tree Detection and Counting on UAVs for Sequential Aerial Images with Generating Orthophoto Mosaicing. Remote Sens, 2022, vol. 14, p. 4113. DOI: 10.3390/rs14164113.
9. Kutyrev A., Andriyanov N., Khort D., Smirnov I., Zubina V. Adaptive CNN Ensemble for Apple Detection: Enabling Sustainable Monitoring Orchard. AgriEngineering, 2025, vol. 7(11), p. 369. DOI: 10.3390/agriengineering7110369.
10. Ali M.L., Zhang Z. The YOLO Framework: A Comprehensive Review of Evolution, Applications, and Benchmarks in Object Detection. Computers, 2024, vol. 13 (12), p. 336. DOI: 10.3390/computers13120336.
11. Lv M., Xu Y.-X., Miao Y.-H., Su W.-H. A Comprehensive Review of Deep Learning in Computer Vision for Monitoring Apple Tree Growth and Fruit Production. Sensors, 2025, vol. 25 (8), p. 2433. DOI: 10.3390/s25082433.
12. Yang Z., Khan Z., Shen Y., Liu H. GTDRYOLOv12: Optimizing YOLO for Efficient and Accurate Weed Detection in Agriculture. Agronomy, 2025, vol. 15 (8), p. 1824. DOI: 10.3390/agronomy15081824.
13. Ribeiro D., Tavares D., Tiradentes E., Santos F., Rodriguez D. Performance Evaluation of YOLOv11 and YOLOv12 Deep Learning Architectures for Automated Detection and Classification of Immature Macauba (Acrocomia aculeata) Fruits. Agriculture, 2025, vol. 15 (15). p. 1571. DOI: 10.3390/agriculture15151571.
14. Padilla R., Passos W.L., Dias T.L.B., Netto S.L., da Silva E.A.B. A Comparative Analysis of Object Detection Metrics with a Companion OpenSource Toolkit. Electronics, 2021, vol. 10 (3), p. 279. DOI: 10.3390/electronics10030279.
15. Hort D.O., Kutyrev A.I., Smirnov I.G., Voronkov I.V. Development of an automated management system for agricultural technologies in horticulture. Sel'skohozyajstvennye mashiny i tekhnologii=Agricultural machinery and technology. 2021, vol. 15 (2), pp. 61–68. (In Russian). DOI: 10.22314/2073-7599-2021-15-2-61-68.
Review
For citations:
Kutyrev A.I. Application of computer vision and deep learning algorithms for tree inventory in intensive orchards. Siberian Herald of Agricultural Science. 2026;56(2):90-100. (In Russ.) https://doi.org/10.26898/0370-8799-2026-2-10
Views:
2439
JATS XML
Email this article 