<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE root>
<article 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="en"><front><journal-meta><journal-id journal-id-type="publisher-id">RUDN Journal of Ecology and Life Safety</journal-id><journal-title-group><journal-title xml:lang="en">RUDN Journal of Ecology and Life Safety</journal-title><trans-title-group xml:lang="ru"><trans-title>Вестник Российского университета дружбы народов. Серия: Экология и безопасность жизнедеятельности</trans-title></trans-title-group></journal-title-group><issn publication-format="print">2313-2310</issn><issn publication-format="electronic">2408-8919</issn><publisher><publisher-name xml:lang="en">Peoples’ Friendship University of Russia named after Patrice Lumumba (RUDN University)</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="publisher-id">52214</article-id><article-id pub-id-type="doi">10.22363/2313-2310-2026-34-3-500-514</article-id><article-id pub-id-type="edn">EVYAWA</article-id><article-categories><subj-group subj-group-type="toc-heading" xml:lang="en"><subject>Environmental Monitoring</subject></subj-group><subj-group subj-group-type="toc-heading" xml:lang="ru"><subject>Экологический мониторинг</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">Experimental multi-criteria evaluation of convolutional neural network architectures for early forest fire detection using UAV data</article-title><trans-title-group xml:lang="ru"><trans-title>Экспериментальная многокритериальная оценка архитектур сверточных нейронных сетей для задачи раннего обнаружения лесных пожаров по данным БПЛА</trans-title></trans-title-group></title-group><contrib-group><contrib contrib-type="author"><contrib-id contrib-id-type="orcid">https://orcid.org/0009-0009-7500-7307</contrib-id><contrib-id contrib-id-type="spin">8839-5769</contrib-id><name-alternatives><name xml:lang="en"><surname>Zagorodnii</surname><given-names>Serafim S.</given-names></name><name xml:lang="ru"><surname>Загородний</surname><given-names>Серафим Сергеевич</given-names></name></name-alternatives><bio xml:lang="en"><p>Postgraduate Student</p></bio><bio xml:lang="ru"><p>аспирант</p></bio><email>zagorodniy.serafim@mail.ru</email><xref ref-type="aff" rid="aff1"/></contrib><contrib contrib-type="author"><contrib-id contrib-id-type="orcid">https://orcid.org/0009-0000-6337-5342</contrib-id><contrib-id contrib-id-type="spin">3560-5659</contrib-id><name-alternatives><name xml:lang="en"><surname>Belova</surname><given-names>Irina K.</given-names></name><name xml:lang="ru"><surname>Белова</surname><given-names>Ирина Константиновна</given-names></name></name-alternatives><bio xml:lang="en"><p>Associate Professor, Candidate of Physical and Mathematical Sciences, Kaluga Branch</p></bio><bio xml:lang="ru"><p>кандидат физико-математических наук, Калужский филиал</p></bio><email>belova.ik@bmstu.ru</email><xref ref-type="aff" rid="aff2"/></contrib><contrib contrib-type="author"><contrib-id contrib-id-type="orcid">https://orcid.org/0009-0001-5958-5168</contrib-id><name-alternatives><name xml:lang="en"><surname>Nikulina</surname><given-names>Svetlana N.</given-names></name><name xml:lang="ru"><surname>Никулина</surname><given-names>Светлана Николаевна</given-names></name></name-alternatives><bio xml:lang="en"><p>Associate Professor, Candidate of Technical Sciences</p></bio><bio xml:lang="ru"><p>кандидат технических наук, доцент</p></bio><email>nikulina-sn@rudn.ru</email><xref ref-type="aff" rid="aff1"/></contrib></contrib-group><aff-alternatives id="aff1"><aff><institution xml:lang="en">RUDN University</institution></aff><aff><institution xml:lang="ru">Российский университет дружбы народов</institution></aff></aff-alternatives><aff-alternatives id="aff2"><aff><institution xml:lang="en">Bauman Moscow State Technical University</institution></aff><aff><institution xml:lang="ru">МГТУ им. Н.Э. Баумана</institution></aff></aff-alternatives><pub-date date-type="pub" iso-8601-date="2026-09-10" publication-format="electronic"><day>10</day><month>09</month><year>2026</year></pub-date><volume>34</volume><issue>3</issue><issue-title xml:lang="en">VOL 34, NO2 (2026)</issue-title><issue-title xml:lang="ru">ТОМ 34, №2 (2026)</issue-title><fpage>500</fpage><lpage>514</lpage><history><date date-type="received" iso-8601-date="2026-09-10"><day>10</day><month>09</month><year>2026</year></date></history><permissions><copyright-statement xml:lang="en">Copyright ©; 2026, Zagorodnii S.S., Belova I.K., Nikulina S.N.</copyright-statement><copyright-statement xml:lang="ru">Copyright ©; 2026, Загородний С.С., Белова И.К., Никулина С.Н.</copyright-statement><copyright-year>2026</copyright-year><copyright-holder xml:lang="en">Zagorodnii S.S., Belova I.K., Nikulina S.N.</copyright-holder><copyright-holder xml:lang="ru">Загородний С.С., Белова И.К., Никулина С.Н.</copyright-holder><ali:free_to_read xmlns:ali="http://www.niso.org/schemas/ali/1.0/"/><license><ali:license_ref xmlns:ali="http://www.niso.org/schemas/ali/1.0/">https://creativecommons.org/licenses/by-nc/4.0</ali:license_ref></license></permissions><self-uri xlink:href="https://journals.rudn.ru/ecology/article/view/52214">https://journals.rudn.ru/ecology/article/view/52214</self-uri><abstract xml:lang="en"><p>The objective of this study is to conduct a comparative analysis of four modern convolutional neural network architectures - ResNet50, MobileNetV2, MobileNetV3 Large, and EfficientNet-B0 - for wildfire detection on multimodal RGBT images captured by UAV-mounted cameras, with emphasis on practical applicability under resource-constrained conditions. The object of study consists of deep learning algorithms trained on the FireMan UAV RGBT dataset, comprising 12,000 annotated RGB and thermal images. The methodology is based on transfer learning with ImageNet-pretrained weights, the use of Focal Loss to address class imbalance, and 5-fold cross-validation. Objective evaluation employed non-parametric statistical tests (Kruskal - Wallis, Mann - Whitney) and a multi-criteria ranking using TOPSIS with bootstrap estimation, considering four key metrics: accuracy (F1-score), sensitivity, inference speed (FPS), and computational complexity (MACs). Experiments were conducted on the NVIDIA Jetson Xavier NX platform. Results showed that all architectures achieve high accuracy (F1-score &gt; 0.92), but MobileNetV2 delivers the optimal balance: F1 = 0.934, 18.2 FPS, and minimal computational load (1.3 G MACs), making it the most suitable for autonomous UAV-based monitoring systems. ResNet50 and EfficientNet-B0 offer comparable accuracy but require 2-3 times more computational resources. MobileNetV3 Large achieved the highest sensitivity but lagged in speed. The scientific novelty lies in a statistically rigorous evaluation of CNN architectures not only by accuracy but also by real-world computational constraints, enabling a transition from laboratory experiments to practical deployment on UAVs. The findings provide a robust basis for selecting optimal architectures in resource-limited environmental monitoring systems.</p></abstract><trans-abstract xml:lang="ru"><p>Цель исследования - сравнительный анализ четырех современных архитектур сверточных нейронных сетей: ResNet50, MobileNetV2, MobileNetV3 Large и EfficientNet-B0 - для детекции лесных пожаров на мультимодальных RGBT-изображениях, снятых с бортовых камер БПЛА, с акцентом на практическую применимость в условиях ограниченных ресурсов. Объектом исследования выступают алгоритмы глубокого обучения, обученные на датасете FireMan UAV RGBT, включающем 12 000 аннотированных изображений в RGB и тепловом диапазонах. Методология основана на трансферном обучении с предобученными весами ImageNet, использовании фокальной функции потерь для компенсации дисбаланса классов и 5-фолд кросс-валидации. Для объективной оценки применены непараметрические статистические тесты (Краскела - Уоллиса, Манна - Уитни) и многокритериальная ранжировка по методу TOPSIS с бутстрэп-оценкой, учитывающей четыре ключевых параметра: точность (F1-score), чувствительность, скорость инференса (FPS) и вычислительную сложность (MACs). Эксперименты проведены на платформе NVIDIA Jetson Xavier NX. Результаты показали, что все архитектуры достигают высокой точности (F1-score &gt; 0,92), однако MobileNetV2 демонстрирует оптимальный баланс: F1 = 0,934, скорость 18,2 FPS при минимальной вычислительной нагрузке (1.3 G MACs), что делает ее наиболее подходящей для автономных систем мониторинга. ResNet50 и EfficientNet-B0 обеспечивают сопоставимую точность, но требуют в 2-3 раза больше ресурсов. MobileNetV3 Large показала наилучшую чувствительность, но уступает по скорости. Научная новизна работы заключается в комплексной, статистически обоснованной оценке CNN-архитектур не только по точности, но и по реальным ограничениям вычислительной эффективности, что позволяет перейти от лабораторных экспериментов к практическому внедрению на БПЛА. Полученные данные формируют основу для выбора оптимальной архитектуры в системах экологического мониторинга с ограниченными ресурсами.</p></trans-abstract><kwd-group xml:lang="en"><kwd>multimodal classification</kwd><kwd>transfer learning</kwd><kwd>MobileNetV2</kwd><kwd>ResNet50</kwd><kwd>EfficientNet B0</kwd><kwd>MobileNetV3 Large</kwd><kwd>focal loss function</kwd></kwd-group><kwd-group xml:lang="ru"><kwd>мультимодальная классификация</kwd><kwd>MobileNetV2</kwd><kwd>ResNet50</kwd><kwd>EfficientNet B0</kwd><kwd>MobileNetV3 Large</kwd><kwd>трансферное обучение</kwd><kwd>фокальная функция потерь</kwd></kwd-group><funding-group/></article-meta><fn-group/></front><body></body><back><ref-list><ref id="B1"><label>1.</label><citation-alternatives><mixed-citation xml:lang="en">Akhloufi MA, Castro NA, Couturier A. UAVs for wildland fires.  Autonomous Systems: Sensors, Vehicles, Security, and the Internet of Everything. 2018;10643:134–147. https://doi.org/10.1117/12.2304834</mixed-citation><mixed-citation xml:lang="ru">Akhloufi M.A., Castro N. A., Couturier A. UAVs for wildland fires // Autonomous Systems: Sensors, Vehicles, Security, and the Internet of Everything. SPIE, 2018. Vol. 10643. P. 134–147. https://doi.org/10.1117/12.2304834</mixed-citation></citation-alternatives></ref><ref id="B2"><label>2.</label><citation-alternatives><mixed-citation xml:lang="en">Bouguettaya A, Zarzour H, Taberkit AM, Kechida A. A review on early wildfire detection from unmanned aerial vehicles using deep learning-based computer vision algorithms. Signal Processing. 2022;190:108309. https://doi.org/10.1016/j.sigpro.2021.108309 EDN: HTSLZA</mixed-citation><mixed-citation xml:lang="ru">Bouguettaya A., Zarzour H., Taberkit A. M., Kechida A. A review on early wildfire detection from unmanned aerial vehicles using deep learning-based computer vision algorithms // Signal Processing. 2022. Vol. 190. Article no. 108309. https://doi.org/10.1016/j.sigpro.2021.108309 EDN: HTSLZA</mixed-citation></citation-alternatives></ref><ref id="B3"><label>3.</label><citation-alternatives><mixed-citation xml:lang="en">Burton C, Lampe S, Kelley DI,  Hantson S, Christidis N, Gudmundsson L, Forrest M, Burke E, Chang J, Huang H, Ito A, Kou-Giesbrecht S, Lasslop G, Li W, Nieradzik L, Li F, Chen Ya, Randerson Ja, Reyer ChPO, Mengel M. Global burned area increasingly explained by climate change. Nature Climate Change. 2024;14(11):1186–1192. https://doi.org/10.1038/s41558-024-02140-w EDN: READTK</mixed-citation><mixed-citation xml:lang="ru">Burton Ch., Lampe S., Kelley D.I., Thiery W., Hantson S., Christidis N., Gudmundsson L., Forrest M., Burke E., Chang J., Huang H., Ito A., Kou-Giesbrecht S., Lasslop G., Li W., Nieradzik L., Li F., Chen Ya., Randerson Ja., Reyer Ch.P.O., Mengel M. Global burned area increasingly explained by climate change // Nature Climate Change. 2024. Vol. 14. No. 11. P. 1186–1192. https://doi.org/10.1038/s41558-024-02140-w EDN: READTK</mixed-citation></citation-alternatives></ref><ref id="B4"><label>4.</label><citation-alternatives><mixed-citation xml:lang="en">Ghali R, Akhloufi MA, Mseddi WS. Deep learning and transformer approaches for UAV-based wildfire detection and segmentation. Sensors. 2022;22(5):1977. https://doi.org/10.3390/s22051977 EDN: HYRIMZ</mixed-citation><mixed-citation xml:lang="ru">Ghali R., Akhloufi M.A., Mseddi W.S. Deep learning and transformer approaches for UAV-based wildfire detection and segmentation // Sensors. 2022. Vol. 22. No. 5. Article no. 1977. https://doi.org/10.3390/s22051977 EDN: HYRIMZ</mixed-citation></citation-alternatives></ref><ref id="B5"><label>5.</label><citation-alternatives><mixed-citation xml:lang="en">Ghali R, Akhloufi MA. Deep learning approaches for wildland fires remote sensing: classification, detection, and segmentation. Remote Sensing. 2023;15(7):1821. https://doi.org/10.3390/rs15071821 EDN: BRAETM</mixed-citation><mixed-citation xml:lang="ru">Ghali R., Akhloufi M.A. Deep learning approaches for wildland fires remote sensing: classification, detection, and segmentation // Remote Sensing. 2023. Vol. 15. No. 7. Article no. 1821. https://doi.org/10.3390/rs15071821 EDN: BRAETM</mixed-citation></citation-alternatives></ref><ref id="B6"><label>6.</label><citation-alternatives><mixed-citation xml:lang="en">Hossain FMA, Zhang YM, Tonima MA. Forest fire flame and smoke detection from UAV-captured images using fire-specific color features and multi-color space local binary pattern. Journal of Unmanned Vehicle Systems. 2020;8(4):285–309. https://doi.org/10.1139/juvs-2020-0009 EDN: JPMXBA</mixed-citation><mixed-citation xml:lang="ru">Hossain F.M.A., Zhang Y.M., Tonima M.A. Forest fire flame and smoke detection from UAV-captured images using fire-specific color features and multi-color space local binary pattern // Journal of Unmanned Vehicle Systems. 2020. Vol. 8. No. 4. P. 285–309. https://doi.org/10.1139/juvs-2020-0009 EDN: JPMXBA</mixed-citation></citation-alternatives></ref><ref id="B7"><label>7.</label><citation-alternatives><mixed-citation xml:lang="en">Khan A, Hassan B, Khan S,  Ahmed R, Abuassba A. DeepFire: A novel dataset and deep transfer learning benchmark for forest fire detection. Mobile Information Systems. 2022;2022(1):5358359. https://doi.org/10.1155/2022/5358359 EDN: PJBAGK</mixed-citation><mixed-citation xml:lang="ru">Khan A., Hassan B., Khan S.,  Ahmed R., Abuassba A. DeepFire: a novel dataset and deep transfer learning benchmark for forest fire detection // Mobile Information Systems. 2022. Vol. 2022. No. 1. Article no. 5358359. https://doi.org/10.1155/2022/5358359 EDN: PJBAGK</mixed-citation></citation-alternatives></ref><ref id="B8"><label>8.</label><citation-alternatives><mixed-citation xml:lang="en">Kularatne W, Sedaghat Shayegan K, Álvarez Casado C., Rajala J,  Hänninen, T, Bordallo López M, Nguyen N Le. FireMan-UAV-RGBT (1.0.0-beta.2). Zenodo. 2024. https://doi.org/10.5281/zenodo.13732947</mixed-citation><mixed-citation xml:lang="ru">Kularatne W., Sedaghat Shayegan K., Álvarez Casado C., Rajala J.,  Hänninen, T., Bordallo López M., Nguyen N. Le. FireMan-UAV-RGBT (1.0.0-beta.2) // Zenodo. 2024. https://doi.org/10.5281/zenodo.13732947</mixed-citation></citation-alternatives></ref><ref id="B9"><label>9.</label><citation-alternatives><mixed-citation xml:lang="en">Perevedentsev Y, Sherstyukov B, Gusarov A, Aukhadeev T., Mirsaeva N. Climate-induced fire hazard in forests in the Volga federal district of European Russia during 1992–2020. Climate. 2022;10(7):110. https://doi.org/10.3390/cli10070110 EDN: WIFGSO</mixed-citation><mixed-citation xml:lang="ru">Perevedentsev Yu, Sherstyukov B, Gusarov A., Aukhadeev T., Mirsaeva N. Climate-induced fire hazard in forests in the Volga federal district of European Russia during 1992–2020 // Climate. 2022. Vol. 10. No. 7. Article no. 110. https://doi.org/10.3390/cli10070110 EDN: WIFGSO</mixed-citation></citation-alternatives></ref><ref id="B10"><label>10.</label><citation-alternatives><mixed-citation xml:lang="en">Pryadilina NK, Zinoveva IS, Skvorcov EA. Economic assessment of the consequences of forest fires (on the example of the Ural federal district). Actual Directions of Scientific Researches of the XXI Century: Theory and Practice. 2024;12(1):79–94. https://doi.org/10.34220/2308-8877-2024-12-1-79-94 EDN: AZXDJX</mixed-citation><mixed-citation xml:lang="ru">Pryadilina N.K., Zinoveva I. S., Skvorcov E.A. Economic assessment of the consequences of forest fires (on the example of the Ural federal district) // Актуальные направления научных исследований XXI века: теория и практика. 2024. Т. 12. № 1. С. 79–94. https://doi.org/10.34220/2308-8877-2024-12-1-79-94 EDN: AZXDJX</mixed-citation></citation-alternatives></ref><ref id="B11"><label>11.</label><citation-alternatives><mixed-citation xml:lang="en">Shamsoshoara A, Afghah F, Razi A. Aerial imagery pile burn detection using deep learning: the FLAME dataset. Computer Networks. 2021;193:108001. https://doi.org/10.1016/j.comnet.2021.108001 EDN: OOWULQ</mixed-citation><mixed-citation xml:lang="ru">Shamsoshoara A., Afghah F., Razi A., Zheng L., Fulé P., Blasch E. Aerial imagery pile burn detection using deep learning: the FLAME dataset // Computer Networks. 2021. Vol. 193. Article no. 108001. https://doi.org/10.1016/j.comnet.2021.108001 EDN: OOWULQ</mixed-citation></citation-alternatives></ref><ref id="B12"><label>12.</label><citation-alternatives><mixed-citation xml:lang="en">Wonjae Lee, Seonghyun Kim, Yong-Tae Lee, Hyun-Woo Lee, Min Choi. Deep neural networks for wild fire detection with unmanned aerial vehicle. 2017 IEEE International Conference on Consumer Electronics (ICCE). Las Vegas, NV; 2017. p. 252–253, https://doi.org/10.1109/ICCE.2017.7889305.</mixed-citation><mixed-citation xml:lang="ru">Wonjae Lee, Seonghyun Kim, Yong-Tae Lee, Hyun-Woo Lee, Min Choi. Deep neural networks for wild fire detection with unmanned aerial vehicle // 2017 IEEE International Conference on Consumer Electronics (ICCE). Las Vegas, NV, 2017. P. 252–253, https://doi.org/10.1109/ICCE.2017.7889305.</mixed-citation></citation-alternatives></ref><ref id="B13"><label>13.</label><citation-alternatives><mixed-citation xml:lang="en">Nikulina SN, Belova IK, Vasiliev VV. Selection of the optimal method for training neural networks for solving environmental monitoring problems of industrial zones. In: Current Problems of Ecology and Environmental Management: Collection of Scientific Papers of the XXVI International Scientific and Practical Conference. In 3 vols. Moscow; 2025. p. 130–138. (In Russ.)</mixed-citation><mixed-citation xml:lang="ru">Никулина С.Н., Белова И.К., Васильев В.В. Выбор оптимального метода обучения нейронных сетей для решения задач экологического мониторинга промышленных зон // Актуальные проблемы экологии и природопользования : сборник научных трудов XXVI Международной научно-практической конференции : в 2 томах. Москва, 25-27 апреля 2025 г. Москва : РУДН, 2025. Т. 2. С. 130–138.</mixed-citation></citation-alternatives></ref><ref id="B14"><label>14.</label><citation-alternatives><mixed-citation xml:lang="en">Vasiliev VV, Belova IK. Data preparation for automatic relief recognition from unmanned aerial vehicles using neural networks. In: Research and Innovation: Synergy of Knowledge and Practice: Collection of Articles of the II International Scientific and Practical Conference. Moscow; 2025. p. 55–61. (In Russ.) EDN: KEJWBU</mixed-citation><mixed-citation xml:lang="ru">Васильев В.В., Белова И.К. Подготовка данных для автоматического распознавания рельефа с беспилотных летательных аппаратов с использованием нейронных сетей // Исследования и инновации: синергия знаний и практики : сборник статей II Международной научно-практической конференции. Москва, 2025. С. 55–61. EDN: KEJWBU</mixed-citation></citation-alternatives></ref><ref id="B15"><label>15.</label><citation-alternatives><mixed-citation xml:lang="en">Belova IK, Vidyaev KA. Using neural network technologies for assessing the condition of forest resources. Paradigma. 2025;(4-2):26–31. (In Russ.) EDN: YEZERT</mixed-citation><mixed-citation xml:lang="ru">Белова И.К., Видяев К.А. Использование нейросетевых технологий для оценки состояния лесных ресурсов // Парадигма. 2025. № 4-2. С. 26–31. EDN: YEZERT</mixed-citation></citation-alternatives></ref></ref-list></back></article>
