<?xml version='1.0' encoding='utf-8'?>
<article xmlns:ns0="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:ali="http://www.niso.org/schemas/ali/1.0/" article-type="research-article" dtd-version="1.2" xml:lang="ru">
 <front>
 <journal-meta>
 <journal-id journal-id-type="publisher-id">dongu-vestnik04.ru</journal-id>
 <journal-title-group>
 <journal-title xml:lang="ru">Вестник Донецкого университета. Серия 04. Технические науки</journal-title>
 <trans-title-group xml:lang="en">
 <trans-title>Vestnik of Donetsk University. Series 04. Technical Sciences</trans-title>
 </trans-title-group>
 </journal-title-group>
 <issn publication-format="electronic">2663-4228</issn>
 </journal-meta>
 <article-meta>
 <article-id pub-id-type="publisher-id">535</article-id>
 <article-id pub-id-type="doi">10.5281/zenodo.20800042</article-id>
 <article-categories>
 <subj-group subj-group-type="toc-heading" xml:lang="en">
 <subject>Articles</subject>
 </subj-group>
 <subj-group subj-group-type="toc-heading" xml:lang="ru">
 <subject>Статьи</subject>
 </subj-group>
 <subj-group subj-group-type="article-type">
 <subject>Research Article</subject>
 </subj-group>
 </article-categories>
 <title-group>
 <article-title xml:lang="en">STUDY OF SELECTIVE AUGMENTATION STRATEGIES FOR SEMANTIC SEGMENTATION OF UAV AERIAL IMAGERY</article-title>
 <trans-title-group xml:lang="ru">
 <trans-title>ИССЛЕДОВАНИЕ СТРАТЕГИЙ СЕЛЕКТИВНОЙ АУГМЕНТАЦИИ ДЛЯ СЕМАНТИЧЕСКОЙ СЕГМЕНТАЦИИ АЭРОФОТОСНИМКОВ БПЛА</trans-title>
 </trans-title-group>
 </title-group>
 <contrib-group>
 <contrib contrib-type="author">
 <contrib-id contrib-id-type="orcid">https://orcid.org/0009-0001-6059-3229</contrib-id>
 <name-alternatives>
 <name xml:lang="en">
 <surname>Pikalev</surname>
 <given-names>IAroslav Sergeevich</given-names>
 </name>
 <name xml:lang="ru">
 <surname>Пикалёв</surname>
 <given-names>Ярослав Сергеевич</given-names>
 </name>
 </name-alternatives>
 <email>i@pikaliov.ru</email>
 <xref ref-type="aff" rid="aff1">1</xref>
 </contrib>
 <contrib contrib-type="author">
 <contrib-id contrib-id-type="orcid">https://orcid.org/0009-0005-0354-9618</contrib-id>
 <name-alternatives>
 <name xml:lang="en">
 <surname>Ustenko</surname>
 <given-names>Vladimir IUrevich</given-names>
 </name>
 <name xml:lang="ru">
 <surname>Устенко</surname>
 <given-names>Владимир Юрьевич</given-names>
 </name>
 </name-alternatives>
 <email>ustencko.vova@ya.ru</email>
 <xref ref-type="aff" rid="aff2">2</xref>
 </contrib>
 </contrib-group>
 <aff-alternatives id="aff1">
 <aff xml:lang="ru">
 <institution>ФГБНУ «Институт проблем искусственного интеллекта»</institution>
 </aff>
 <aff xml:lang="en">
 <institution>Institute of Artificial Intelligence Problems</institution>
 </aff>
 </aff-alternatives>
 <aff-alternatives id="aff2">
 <aff xml:lang="ru">
 <institution>ФГБНУ «Институт проблем искусственного интеллекта»</institution>
 </aff>
 <aff xml:lang="en">
 <institution>nstitute of Artificial Intelligence Problems</institution>
 </aff>
 </aff-alternatives>
 <pub-date date-type="pub" iso-8601-date="05.06.2026" publication-format="electronic" />
 <issue>2</issue>
 <issue-title xml:lang="en">NO2 (05.0)</issue-title>
 <issue-title xml:lang="ru">№2 (05.0)</issue-title>
 <fpage>127</fpage>
 <lpage>140</lpage>
 <history>
 <date date-type="received" iso-8601-date="2026-07-23">
 <day>23</day>
 <month>07</month>
 <year>2026</year>
 </date>
 </history>
 <permissions>
 <copyright-statement xml:lang="ru">Copyright ©; 2026, Вестник Донецкого университета. Серия 04. Технические науки</copyright-statement>
 <copyright-statement xml:lang="en">Copyright ©; 2026, Vestnik of Donetsk University. Series 04. Technical Sciences</copyright-statement>
 <copyright-year>2026</copyright-year>
 <copyright-holder xml:lang="ru">Вестник Донецкого университета. Серия 04. Технические науки</copyright-holder>
 <copyright-holder xml:lang="en">Vestnik of Donetsk University. Series 04. Technical Sciences</copyright-holder>
 <license license-type="open-access" ns0:href="https://creativecommons.org/licenses/by-nc/4.0/" xml:lang="ru">
 <license-p>Эта статья распространяется на условиях лицензии Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)</license-p>
 </license>
 <license license-type="open-access" ns0:href="https://creativecommons.org/licenses/by-nc/4.0/" xml:lang="en">
 <license-p>This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0)</license-p>
 </license>
 <ali:free_to_read />
 </permissions>
 <self-uri ns0:href="https://dongu-vestnik04.ru/index.php/vestnik04/article/view/535">https://dongu-vestnik04.ru/index.php/vestnik04/article/view/535</self-uri>
 <abstract xml:lang="en">
 <p>Semantic segmentation of aerial images captured by unmanned aerial vehicles (UAVs) is used for urban infrastructure monitoring. Class imbalance remains an unresolved problem during training: small-sized infrastructure objects occupy less than two percent of the scene pixels, resulting in insufficient segmentation accuracy for these categories. Existing data augmentation methods address this problem with reasonable effectiveness. The RareClassToGrayAugMask method, proposed in this paper, performs selective desaturation of pixels only within the masks of designated semantic classes and additionally distorts the desaturation region, which improves robustness and model invariance to color in the areas of rare objects.</p>
 </abstract>
 <trans-abstract xml:lang="ru">
 <p>Семантическая сегментация аэрофотоснимков полученных с помощью беспилотных летательных аппаратов (БПЛА) применяется для мониторинга городской инфраструктуры. При обучении дисбаланс классов остаётся нерешённой проблемой: объекты инфраструктуры малого размера занимают менее двух процентов пикселей сцены, что приводит к недостаточной точности сегментации на этих категориях. Существующие методы расширения данных довольно эффективно позволяют бороться с проблемой. Метод RareClassToGrayAugMask, предложенный в настоящей статье, выполняет селективное обесцвечивание пикселей только в областях масок заданных семантических классов, а также искажает область обесцвечивания, что повышает робастность, инвариантность модели к цвету в областях редких объектов.</p>
 </trans-abstract>
 <kwd-group xml:lang="en">
 <kwd>semantic segmentation, StripNet, FPN, SynDrone, UAV, aerial imagery, selective augmentation, RareClassToGrayAugMask, Copy-Paste</kwd>
 </kwd-group>
 <kwd-group xml:lang="ru">
 <kwd>семантическая сегментация, StripNet, FPN, SynDrone, БПЛА, аэрофотосъёмка, селективная аугментация, RareClassToGrayAugMask, Copy-Paste</kwd>
 </kwd-group>
 <funding-group>
 <funding-statement xml:lang="en">Результаты исследований получены в рамках НИР FREN-2024-0002 «Извлечение семантической информации из изображений для автономных систем навигации БПЛА».</funding-statement>
 <funding-statement xml:lang="ru">Результаты исследований получены в рамках НИР FREN-2024-0002 «Извлечение семантической информации из изображений для автономных систем навигации БПЛА».</funding-statement>
 </funding-group>
 </article-meta>
 </front>
 <body>
 <p>[Полный текст статьи отсутствует в исходных данных. Необходимо добавить текст из PDF или другого источника.]</p>
 </body>
 <back>
 <ref-list>
 <title>Список литературы</title>
 <ref id="B1">
 <mixed-citation>1. Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation / L.-C. Chen, Y. Zhu, G. Papandreou [et al.] // Lecture Notes in Computer Science. – 2018. – Vol. 11211. – P. 801-818.</mixed-citation>
 </ref>
 <ref id="B2">
 <mixed-citation>2. Зуев, В. М. Сравнение обнаружения объектов средствами искусственного интеллекта в сравнении с классическими методами / В. М. Зуев // Проблемы искусственного интеллекта. – 2024. – Т. 34, № 3. – С. 4-10.</mixed-citation>
 </ref>
 <ref id="B3">
 <mixed-citation>3. SynDrone – Multi-modal UAV Dataset for Urban Scenarios / G. Rizzoli, F. Barbato, M. Caligiuri [et al.] // Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). – 2023. – P. 2236-2246.</mixed-citation>
 </ref>
 <ref id="B4">
 <mixed-citation>4. Пикалёв, Я. С. Обнаружение ключевых объектов и перекрёстная геолокализация: анализ наборов данных и методологические перспективы / Я. С. Пикалёв // Проблемы искусственного интеллекта. – 2024. – № 4 (35). – С. 25-37. – DOI 10.24412/2413-7383-2024-4-25-37.</mixed-citation>
 </ref>
 <ref id="B5">
 <mixed-citation>5. Устенко, В. Ю. Методика расширения данных для сегментации ключевых объектов на аэрофотоснимках БПЛА / В. Ю. Устенко, М. В. Близно // Проблемы искусственного интеллекта. – 2025. – № 4 (39). – С. 84-97. – DOI 10.24412/2413-7383-2025-4-39-84-97.</mixed-citation>
 </ref>
 <ref id="B6">
 <mixed-citation>6. Image Segmentation Using Deep Learning: A Survey / S. Minaee, Y. Boykov, F. Porikli [et al.] // IEEE Transactions on Pattern Analysis and Machine Intelligence. – 2022. – Vol. 44, no. 7. – P. 3523-3542.</mixed-citation>
 </ref>
 <ref id="B7">
 <mixed-citation>7. Shorten, C. A Survey on Image Data Augmentation for Deep Learning / C. Shorten, T. M. Khoshgoftaar // Journal of Big Data. – 2019. – Vol. 6, No. 1. – Art. 60.</mixed-citation>
 </ref>
 <ref id="B8">
 <mixed-citation>8. DeVries, T. Improved Regularization of Convolutional Neural Networks with Cutout / T. DeVries, G. W. Taylor // arXiv preprint. – 2017. – URL: https://arxiv.org/abs/1708.04552.</mixed-citation>
 </ref>
 <ref id="B9">
 <mixed-citation>9. Masked Autoencoders Are Scalable Vision Learners / K. He, X. Chen, S. Xie [et al.] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). – 2022. – P. 16000-16009.</mixed-citation>
 </ref>
 <ref id="B10">
 <mixed-citation>10. Choi, J. Colorful Cutout: Enhancing Image Data Augmentation with Curriculum Learning / J. Choi, Y. Kim // arXiv preprint. – 2024. – URL: https://arxiv.org/pdf/2403.20012.</mixed-citation>
 </ref>
 <ref id="B11">
 <mixed-citation>11. Mixup: Beyond empirical risk minimization/ H. Zhang, M. Cisse, Y. N. Dauphin [et al.] // arXiv preprint. – 2017. – URL: https://arxiv.org/abs/1710.09412.</mixed-citation>
 </ref>
 <ref id="B12">
 <mixed-citation>12. CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features / S. Yun, D. Han, S. Chun [et al.] // Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). – 2019. – P. 6023-6032.</mixed-citation>
 </ref>
 <ref id="B13">
 <mixed-citation>13. Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation / G. Ghiasi, Y. Cui, A. Srinivas [et al.] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). – 2021. – P. 2917-2928.</mixed-citation>
 </ref>
 <ref id="B14">
 <mixed-citation>14. ImageNet-trained CNNs are biased towards textures; increasing shape bias improves accuracy and robustness / R. Geirhos, P. Rubisch, C. Michaelis [et al.] // Proceedings of the International Conference on Learning Representations (ICLR). – 2019. – 22 p.</mixed-citation>
 </ref>
 <ref id="B15">
 <mixed-citation>15. Lee, J. Improving Robustness to Texture Bias via Shape-Focused Augmentation / J. Lee, S. Kim, S. Kim [et al.] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). – 2022. – P. 4323-4331</mixed-citation>
 </ref>
 <ref id="B16">
 <mixed-citation>16. Deep Long-Tailed Learning: A Survey / Y. Zhang, B. Kang, B. Hooi [et al.] // IEEE Transactions on Pattern Analysis and Machine Intelligence. – 2023. – Vol. 45, No. 9. – P. 10795-10816.</mixed-citation>
 </ref>
 <ref id="B17">
 <mixed-citation>17. CP2M: Clustered-Patch-Mixed Mosaic Augmentation for Aerial Image Segmentation / Y. Li [et al.] // arXiv preprint. – 2025. – URL: https://arxiv.org/pdf/2501.15389.</mixed-citation>
 </ref>
 <ref id="B18">
 <mixed-citation>18. UAVid: A Semantic Segmentation Dataset for UAV Imagery / Y. Lyu, G. Vosselman, G.-S. Xia [et al.] // ISPRS Journal of Photogrammetry and Remote Sensing. – 2020. – Vol. 165. – P. 108-119. – DOI 10.1016/j.isprsjprs.2020.05.009.</mixed-citation>
 </ref>
 <ref id="B19">
 <mixed-citation>19. Pyramid Scene Parsing Network / H. Zhao, J. Shi, X. Qi [et al.] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). – 2017. – P. 6230-6239.</mixed-citation>
 </ref>
 <ref id="B20">
 <mixed-citation>20. The Pascal Visual Object Classes (VOC) Challenge / M. Everingham, L. Van Gool, C. K. I. Williams [et al.] // International Journal of Computer Vision. – 2010. – Vol. 88, No. 2. – P. 303-338.</mixed-citation>
 </ref>
 <ref id="B21">
 <mixed-citation>21. Hermann, K. L. The Origins and Prevalence of Texture Bias in Convolutional Neural Networks / K. L. Hermann, T. Chen, S. Kornblith // Advances in Neural Information Processing Systems (NeurIPS). – 2020. – Vol. 33. – P. 19000-19015.</mixed-citation>
 </ref>
 <ref id="B22">
 <mixed-citation>22. Feature Pyramid Networks for Object Detection / T.-Y. Lin, P. Dollar, R. Girshick [et al.] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). – 2017. – P. 936-944.</mixed-citation>
 </ref>
 <ref id="B23">
 <mixed-citation>REFERENCES LIST</mixed-citation>
 </ref>
 <ref id="B24">
 <mixed-citation>1. Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation / L.-C. Chen, Y. Zhu, G. Papandreou [et al.] // Lecture Notes in Computer Science. – 2018. – Vol. 11211. – P. 801-818.</mixed-citation>
 </ref>
 <ref id="B25">
 <mixed-citation>2. Zuev, V. M. Sravnenie obnaruzheniia obieektov sredstvami iskusstvennogo intellekta v sravnenii s klassicheskimi metodami / V. M. Zuev // Problemy iskusstvennogo intellekta. – 2024. – T. 34, № 3. – S. 4-10.</mixed-citation>
 </ref>
 <ref id="B26">
 <mixed-citation>3. SynDrone – Multi-modal UAV Dataset for Urban Scenarios / G. Rizzoli, F. Barbato, M. Caligiuri [et al.] // Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). – 2023. – P. 2236-2246.</mixed-citation>
 </ref>
 <ref id="B27">
 <mixed-citation>4. Pikalev, Ia. S. Obnaruzhenie kliuchevykh obieektov i perekrestnaia geolokalizatsiia: analiz naborov dannykh i metodologicheskie perspektivy / Ia. S. Pikalev // Problemy iskusstvennogo intellekta. – 2024. – № 4 (35). – S. 25-37. – DOI 10.24412/2413-7383-2024-4-25-37.</mixed-citation>
 </ref>
 <ref id="B28">
 <mixed-citation>5. Ustenko, V. Iu. Metodika rasshireniia dannykh dlia segmentatsii kliuchevykh obieektov na aerofotosnimkakh BPLA / V. Iu. Ustenko, M. V. Blizno // Problemy iskusstvennogo intellekta. – 2025. – № 4 (39). – S. 84-97. – DOI 10.24412/2413-7383-2025-4-39-84-97.</mixed-citation>
 </ref>
 <ref id="B29">
 <mixed-citation>6. Image Segmentation Using Deep Learning: A Survey / S. Minaee, Y. Boykov, F. Porikli [et al.] // IEEE Transactions on Pattern Analysis and Machine Intelligence. – 2022. – Vol. 44, no. 7. – P. 3523-3542.</mixed-citation>
 </ref>
 <ref id="B30">
 <mixed-citation>7. Shorten, C. A Survey on Image Data Augmentation for Deep Learning / C. Shorten, T. M. Khoshgoftaar // Journal of Big Data. – 2019. – Vol. 6, No. 1. – Art. 60.</mixed-citation>
 </ref>
 <ref id="B31">
 <mixed-citation>8. DeVries, T. Improved Regularization of Convolutional Neural Networks with Cutout / T. DeVries, G. W. Taylor // arXiv preprint. – 2017. – URL: https://arxiv.org/abs/1708.04552.</mixed-citation>
 </ref>
 <ref id="B32">
 <mixed-citation>9. Masked Autoencoders Are Scalable Vision Learners / K. He, X. Chen, S. Xie [et al.] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). – 2022. – P. 16000-16009.</mixed-citation>
 </ref>
 <ref id="B33">
 <mixed-citation>10. Choi, J. Colorful Cutout: Enhancing Image Data Augmentation with Curriculum Learning / J. Choi, Y. Kim // arXiv preprint. – 2024. – URL: https://arxiv.org/pdf/2403.20012.</mixed-citation>
 </ref>
 <ref id="B34">
 <mixed-citation>11. Mixup: Beyond empirical risk minimization/ H. Zhang, M. Cisse, Y. N. Dauphin [et al.] // arXiv preprint. – 2017. – URL: https://arxiv.org/abs/1710.09412.</mixed-citation>
 </ref>
 <ref id="B35">
 <mixed-citation>12. CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features / S. Yun, D. Han, S. Chun [et al.] // Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). – 2019. – P. 6023-6032.</mixed-citation>
 </ref>
 <ref id="B36">
 <mixed-citation>13. Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation / G. Ghiasi, Y. Cui, A. Srinivas [et al.] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). – 2021. – P. 2917-2928.</mixed-citation>
 </ref>
 <ref id="B37">
 <mixed-citation>14. ImageNet-trained CNNs are biased towards textures; increasing shape bias improves accuracy and robustness / R. Geirhos, P. Rubisch, C. Michaelis [et al.] // Proceedings of the International Conference on Learning Representations (ICLR). – 2019. – 22 p.</mixed-citation>
 </ref>
 <ref id="B38">
 <mixed-citation>15. Lee, J. Improving Robustness to Texture Bias via Shape-Focused Augmentation / J. Lee, S. Kim, S. Kim [et al.] // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). – 2022. – P. 4323-4331</mixed-citation>
 </ref>
 <ref id="B39">
 <mixed-citation>16. Deep Long-Tailed Learning: A Survey / Y. Zhang, B. Kang, B. Hooi [et al.] // IEEE Transactions on Pattern Analysis and Machine Intelligence. – 2023. – Vol. 45, No. 9. – P. 10795-10816.</mixed-citation>
 </ref>
 <ref id="B40">
 <mixed-citation>17. CP2M: Clustered-Patch-Mixed Mosaic Augmentation for Aerial Image Segmentation / Y. Li [et al.] // arXiv preprint. – 2025. – URL: https://arxiv.org/pdf/2501.15389.</mixed-citation>
 </ref>
 <ref id="B41">
 <mixed-citation>18. UAVid: A Semantic Segmentation Dataset for UAV Imagery / Y. Lyu, G. Vosselman, G.-S. Xia [et al.] // ISPRS Journal of Photogrammetry and Remote Sensing. – 2020. – Vol. 165. – P. 108-119. – DOI 10.1016/j.isprsjprs.2020.05.009.</mixed-citation>
 </ref>
 <ref id="B42">
 <mixed-citation>19. Pyramid Scene Parsing Network / H. Zhao, J. Shi, X. Qi [et al.] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). – 2017. – P. 6230-6239.</mixed-citation>
 </ref>
 <ref id="B43">
 <mixed-citation>20. The Pascal Visual Object Classes (VOC) Challenge / M. Everingham, L. Van Gool, C. K. I. Williams [et al.] // International Journal of Computer Vision. – 2010. – Vol. 88, No. 2. – P. 303-338.</mixed-citation>
 </ref>
 <ref id="B44">
 <mixed-citation>21. Hermann, K. L. The Origins and Prevalence of Texture Bias in Convolutional Neural Networks / K. L. Hermann, T. Chen, S. Kornblith // Advances in Neural Information Processing Systems (NeurIPS). – 2020. – Vol. 33. – P. 19000-19015.</mixed-citation>
 </ref>
 <ref id="B45">
 <mixed-citation>22. Feature Pyramid Networks for Object Detection / T.-Y. Lin, P. Dollar, R. Girshick [et al.] // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). – 2017. – P. 936-944.</mixed-citation>
 </ref>
 </ref-list>
 </back>
 </article>
