УДК 528.8 Дистанционное зондирование. Методы и приборы
ГРНТИ 28.23 Искусственный интеллект
Выборки из аэрокосмических изображений и масок, используемые при решении задач распознавания различных классов земной поверхности, могут оказывать существенное влияние на обучаемость моделей нейронных сетей и получаемые в дальнейшем с их помощью результаты распознавания. Состав выборок данных в большинстве случаев рассматривается не относительно самих выборок, а с точки зрения обработки данных нейронными сетями в целом в конкретной задаче. В контексте семантической сегментации геополей сформулированы общие для задач семантической сегментации объектов на аэрокосмических изображениях проблемы: разные яркостные характеристики снимков, тени, эквивалентность яркостных характеристик объектов целевого класса и других объектов сцен, некорректная разметка, граничные случаи дисбаланса классов. Все перечисленное рассматривается как проблемы представления исходного множества геополей в выборках данных. В результате эксперимента с нейронными сетями U Net, STT и MF‑CNN определено, что включаемые в выборки граничные случаи дисбаланса классов и применение снимков с разрешением, при котором дисбаланс классов выше, чем при использовании частей снимков, существенно снижают обучаемость нейронных сетей и точность распознавания, а отбор данных на основе удаления граничных случаев дисбаланса классов при предобработке позволяет как повысить точность распознавания, так и снизить необходимые для обучения моделей временные затраты.
геополе, множество геополей, семантическая сегментация, состав выборок данных, дисбаланс классов, нейронная сеть, точность распознавания
1. Xu R., Mao R., Zhuang Z., et al. Building Extraction from Remote Sensing Images Based on Multi-Scale Attention Gate and Enhanced Positional Information // PeerJ Computer Science. 2025. Vol. 11. P. e2826. DOIhttps://doi.org/10.7717/peerj-cs.2826. https://doi.org/10.7717/peerj-cs.2826
2. Chen Y., Xie Y., Yao W., et al. U-MGA: a Multi-Module Unet Optimized with Multi-Scale Global Attention Mechanisms for Fine-Grained Segmentation of Cultivated Areas // Remote Sensing. 2025. Vol. 17. Iss. 5. P. 760. DOIhttps://doi.org/10.3390/rs17050760. https://doi.org/10.3390/rs17050760
3. Cai J., Tao L., Li Y. CM-UNet++: A Multi-Level Information Optimized Network for Urban Water Body Extraction from High-Resolution Remote Sensing Imagery // Remote Sensing. 2025. Vol. 17. Iss. 6. P. 980. DOIhttps://doi.org/10.3390/rs17060980. https://doi.org/10.3390/rs17060980
4. Gui L., Gu X., Huang F., et al. Road Extraction from Remote Sensing Images Using a Skip-Connected Parallel CNN-Transformer Encoder-Decoder Model // Applied Sciences. 2025. Vol. 15. Iss. 3. P. 1427. DOIhttps://doi.org/10.3390/app15031427. https://doi.org/10.3390/app15031427
5. Liu Y., Duan Y., Zhang X., et al. FEPA-Net: A Building Extraction Network Based on Fusing the Feature Extraction and Position Attention Module // Applied Sciences. 2025. Vol. 15. Iss. 8. P. 4432. DOIhttps://doi.org/10.3390/app15084432. https://doi.org/10.3390/app15084432
6. Zhu B., Yu D., Xiao X., et al. AP-Pointrend: An Improved Network for Building Extraction via High-Resolution Remote Sensing Images // Remote Sensing. 2025. Vol. 17. Iss. 9. P. 1481. DOIhttps://doi.org/10.3390/rs17091481. https://doi.org/10.3390/rs17091481
7. Li X., Yang S., Meng F., et al. LCMorph: Exploiting Frequency Cues and Morphological Perception for Low-Contrast Road Extraction in Remote Sensing Images // Remote Sensing. 2025. Vol. 17. Iss. 2. P. 257. DOIhttps://doi.org/10.3390/rs17020257. https://doi.org/10.3390/rs17020257
8. Weng Z., Li Q., Zheng Z., et al. SCR-Net: A Dual-Channel Water Body Extraction Model Based on Multi-Spectral Remote Sensing Imagery – A Case Study of Daihai Lake, China // Sensors. 2025. Vol. 25. Iss. 3. P. 763. DOI:https://doi.org/10.3390/s25030763. https://doi.org/10.3390/s25030763
9. Wu Q., Chen M., Shi H., et al. Algorithm for Detecting Trees Affected by Pine Wilt Disease in Complex Scenes Based on CNN-Transformer // Forests. 2025. Vol. 16. Iss. 4. P. 596. DOIhttps://doi.org/10.3390/f16040596. https://doi.org/10.3390/f16040596
10. Wang L., Gao Y., Liu Y., et al. Monitoring Pine Shoot Beetle Damage Using UAV Imagery and Deep Learning Semantic Segmentation Under Different Forest Backgrounds // Forests. 2025. Vol. 16. Iss. 4. P. 668. DOIhttps://doi.org/10.3390/f16040668. https://doi.org/10.3390/f16040668
11. Бирюков Н.А., Майоров А.А., Лапчинская М.П. Семантическая сегментация геополей с использованием нейронных сетей на примере проблематики выделения зданий на космо- и аэрофотоснимках // Известия вузов «Геодезия и аэрофотосъемка». 2024. Т. 68, № 1. С. 44–61. DOIhttps://doi.org/10.30533/GiA-2024-004. https://miigaik.ru/journal/archive/2024/2024_68_1_RU/GiA-2024-004.pdf
12. Li J., He W., Cao W., et al. UANet: An Uncertainty-Aware Network for Building Extraction from Remote Sensing Images // IEEE Transactions on Geoscience and Remote Sensing. 2024. Vol. 62. P. 5608513. DOIhttps://doi.org/10.1109/TGRS.2024.3361211. https://ieeexplore.ieee.org/document/10418227
13. Han R., Fan X., Liu J. EUNet: Edge-Unet for Accurate Building Extraction and Edge Emphasis in Gaofen-7 Images // Remote Sensing. 2024. Vol. 16. Iss. 13. P. 2397. DOIhttps://doi.org/10.3390/rs16132397. https://doi.org/10.3390/rs16132397
14. Fenglei W., Xin G., Zongze Z., et al. A Boundary-Enhanced Semantic Segmentation Model for Buildings // IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 2025. Vol. 18. P. 5733–5748. DOIhttps://doi.org/10.1109/JSTARS.2025.3529456. https://ieeexplore.ieee.org/document/10840290
15. Chen P., Huang H., Ye F., et al. A Benchmark Gaofen-7 Dataset for Building Extraction from Satellite Images // Scientific Data. 2024. Vol. 11. P. 187. DOIhttps://doi.org/10.1038/s41597-024-03009-5. https://doi.org/10.1038/s41597-024-03009-5
16. Li S., Bao T., Liu H., et al. Multilevel Feature Aggregated Network with Instance Contrastive Learning Constraint for Building Extraction // Remote Sensing. 2023. Vol. 15. Iss. 10. P. 2585. DOIhttps://doi.org/10.3390/rs15102585. https://doi.org/10.3390/rs15102585
17. Lyu X., Jiang W., Li X., et al. MSAFNet: Multiscale Successive Attention Fusion Network for Water Body Extraction of Remote Sensing Images // Remote Sensing. 2023. Vol. 15. Iss. 12. P. 3121. DOIhttps://doi.org/10.3390/rs15123121. https://doi.org/10.3390/rs15123121
18. Li M., Hong L., Guo J., et al. Automated Extraction of Lake Water Bodies in Complex Geographical Environments by Fusing Sentinel-1/2 Data // Water. 2022. Vol. 14. Iss. 1. P. 30. DOIhttps://doi.org/10.3390/w14010030. https://doi.org/10.3390/w14010030
19. Wang Y., Li S., Lin Y., et al. Lightweight Deep Neural Network Method for Water Body Extraction from High-Resolution Remote Sensing Images with Multisensors // Sensors. 2021. Vol. 21. Iss. 21. 7397. DOIhttps://doi.org/10.3390/s21217397. https://doi.org/10.3390/s21217397
20. Wang B., Chen Z., Wu L., et al. SADA-Net: A Shape Feature Optimization and Multiscale Context Information-Based Water Body Extraction Method for High-Resolution Remote Sensing Images // IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 2022. Vol. 15. P. 1744–1759. DOIhttps://doi.org/10.1109/JSTARS.2022.3146275. https://ieeexplore.ieee.org/document/9695297
21. Weng Y., Li Z., Tang G., et al. OCNet-Based Water Body Extraction from Remote Sensing Images // Water. 2023. Vol. 15. Iss. 20. P. 3557. DOI:10.3390/ w15203557. https://doi.org/10.3390/w15203557
22. Yu J., Cai Y., Lyu X., et al. Boundary-Guided Semantic Context Network for Water Body Extraction from Remote Sensing Images // Remote Sensing. 2023. Vol. 15. Iss. 17. P. 4325. DOIhttps://doi.org/10.3390/rs15174325. https://doi.org/10.3390/rs15174325
23. Li M., Wu P., Wang B., et al. A Deep Learning Method of Water Body Extraction from High Resolution Remote Sensing Images with Multisensors // IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 2021. Vol. 14. P. 3120–3132. DOIhttps://doi.org/10.1109/JSTARS.2021.3060769. https://ieeexplore.ieee.org/document/9360447
24. Zhao Z., Yang J., Wang M., et al. The PCA-NDWI Urban Water Extraction Model Based on Hyperspectral Remote Sensing // Water. 2024. Vol. 16. Iss. 7. P. 963. DOIhttps://doi.org/10.3390/w16070963. https://doi.org/10.3390/w16070963
25. Sandum H.N., Ørka H.O., Tomic O., et al. Semantic Segmentation of Forest Stands Using Deep Learning // Preprint arXiv.org, 2025. [Электронный ресурс]. Режим доступа: https://arxiv.org/pdf/2504.02471 (дата обращения: 09.06.2025).
26. Lin N., Quan H., He J., et al. Urban Vegetation Extraction from High-Resolution Remote Sensing Imagery on SD-Unet and Vegetation Spectral Features // Remote Sensing. 2023. Vol. 15. Iss. 18. P. 4488. DOIhttps://doi.org/10.3390/rs15184488. https://doi.org/10.3390/rs15184488
27. Chen P., Li X., Peng Y., et al. WSSGCN: Hyperspectral Forest Image Classification via Watershed Superpixel Segmentation and Sparse Graph Convolutional Networks // Forests. 2025. Vol. 16. Iss. 5. P. 827. DOIhttps://doi.org/10.3390/f16050827. https://doi.org/10.3390/f16050827
28. Chen S., Zhang M., Lei F. Mapping Vegetation Types by Different Fully Convolutional Neural Network Structures with Inadequate Training Labels in Complex Landscape Urban Areas // Forests. 2023. Vol. 14. Iss. 9. P. 1768. DOIhttps://doi.org/10.3390/f14091788. https://doi.org/10.3390/f14091788
29. Li Y., Min S., Song B., et al. Multisource High-Resolution Remote Sensing Image Vegetation Extraction with Comprehensive Multifeature Perception // Remote Sensing. 2024. Vol. 16. Iss. 4. P. 712. DOIhttps://doi.org/10.3390/rs16040712. https://doi.org/10.3390/rs16040712
30. Wang B., Yao Y. Mountain Vegetation Classification Method Based on Multi-Channel Semantic Segmentation Model // Remote Sensing. 2024. Vol. 16. Iss. 2. P. 256. DOIhttps://doi.org/10.3390/rs16020256. https://doi.org/10.3390/rs16020256
31. Tao J., Chen Z., Sun Z., et al. Seg-Road: A Segmentation Network for Road Extraction Based on Transformer and CNN with Connectivity Structures // Remote Sensing. 2023. Vol. 15. Iss. 6. P. 1602. DOIhttps://doi.org/10.3390/rs15061602. https://doi.org/10.3390/rs15061602
32. Lin S., Yao X., Liu X., et al. MS-AGAN: Road Extraction via Multi-Scale Information Fusion and Asymmetric Generative Adversarial Networks from High-Resolution Remote Sensing Images under Complex Backgrounds // Remote Sensing. 2023. Vol. 15. Iss. 13. P. 3367. DOIhttps://doi.org/10.3390/rs15133367. https://doi.org/10.3390/rs15133367
33. Zhong B., Dan H., Liu M., et al. FERDNet: High-Resolution Remote Sensing Road Extraction Network Based on Feature Enhancement of Road Directionality // Remote Sensing. 2025. Vol. 17. Iss. 3. P. 376. DOIhttps://doi.org/10.3390/rs17030376. https://doi.org/10.3390/rs17030376
34. Chen J., Yang L., Wang H., et al. Road Extraction from High-Resolution Remote Sensing Images via Local and Global Context Reasoning // Remote Sensing. 2023. Vol. 15. Iss. 17. P. 4177. DOIhttps://doi.org/10.3390/rs15174177. https://doi.org/10.3390/rs15174177
35. Mahara A., Khan M.R.K., Deng L., et al. Automated Road Extraction from Satellite Imagery Integrating Dense Depthwise Dilated Separable Spatial Pyramid Pooling with DeepLabV3+ // Applied Sciences. 2025. Vol. 15. Iss. 3. P. 1027. DOIhttps://doi.org/10.3390/app15031027. https://doi.org/10.3390/app15031027
36. Li B., Tang X., Xiao R., et al. Dual Convolutional Network Based on Hypergraph and Multilevel Feature Fusion for Road Extraction from High-Resolution Remote Sensing Images // International Journal of Digital Earth. 2024. Vol. 17. No. 1. P. 2303354. DOIhttps://doi.org/10.1080/17538947.2024.2303354. https://doi.org/10.1080/17538947.2024.2303354
37. Zhao S., Feng Z., Chen L., et al. DANet: A Semantic Segmentation Network for Remote Sensing of Roads Based on Dual-ASPP Structure // Electronics. 2023. Vol. 12. Iss. 15. P. 3243. DOIhttps://doi.org/10.3390/electronics12153243. https://doi.org/10.3390/electronics12153243
38. Zhao L., Zhang J., Meng X., et al. Road Extraction Method of Remote Sensing Image Based on Deformable Attention Transformer // Symmetry. 2024. Vol. 16. Iss. 4. P. 468. DOIhttps://doi.org/10.3390/sym16040468. https://doi.org/10.3390/sym16040468
39. Zhang Y., Zhang L., Wang Y., et al. AGF-Net: Adaptive Global Feature Fusion Network for Road Extraction from Remote-Sensing Images // Complex & Intelligent Systems. 2024. Vol. 10. P. 4311–4328. DOIhttps://doi.org/10.1007/s40747-024-01364-9. https://doi.org/10.1007/s40747-024-01364-9
40. Ma D., Jiang L., Li J., et al. Water Index and Swin Transformer (WISTE) for Water Body Extraction from Multispectral Remote Sensing Images // GIScience & Remote Sensing. 2023. Vol. 60. No. 1. P. 2251704. DOIhttps://doi.org/10.1080/15481603.2023.2251704. https://doi.org/10.1080/15481603.2023.2251704
