dsaitЦифровые решения и технологии искусственного интеллектаDigital Solutions and Artificial Intelligence Technologies3033-7097Финансовый университет при Правительстве Российской Федерацииdsait-11Research ArticleМАТЕМАТИЧЕСКОЕ МОДЕЛИРОВАНИЕ, ЧИСЛЕНЫЕ МЕТОДЫ И КОМПЛЕКСЫ ПРОГРАММMATHEMATICAL MODELING, NUMERICAL METHODS AND SOFTWARE PACKAGESИсследование соотношения точности и производительности моделей YOLO v8 на специальном наборе данных LacmusBenchmark of YOLO v8 with Lacmus datasethttps://orcid.org/0000-0002-5167-2689ЛабинцевaА. И.LabintsevA. I.

Андрей Иванович Лабинцев - кандидат технических наук, доцент кафедры искусственного интеллекта факультета информационных технологий и анализа больших данных; ведущий аналитик департамента стратегических проектов

Москва

Andrey I. Labintsev - Cand. Sci (Tech.), Assoc. Professor at the Department of Artificial Intelligence, Faculty of Information Technology and Big Data Analysis; Leading Analyst at the Department of Strategic Projects

Moscow

ailabintsev@fa.ruhttps://orcid.org/0000-0002-7312-4763КубликЕ. И.KublikE. I.

Евгений Ильич Кублик - кандидат технических наук, доцент кафедры информационных технологий факультета информационных технологий и анализа больших данных

Москва

Evgeny I. Kublik - Cand. Sci (Tech.), Assoc. Prof. at the Department of Information Technology, Faculty of Information Technology and Big Data Analysis

Moscow

eikublik@fa.ruhttps://orcid.org/0009-0009-7176-6242ПеревозчиковГ. П.PerevozchikovG. P.

Георгий Павлович Перевозчиков - научный сотрудник

Georgiy P. Perevozchikov - Researcher 

Würzburg

gosha20777@live.ruhttps://orcid.org/0000-0003-3186-3901КочкаровР. А.KоchkarovR. A.

Расул Ахматович Кочкаров - кандидат экономических наук, доцент кафедры искусственного интеллекта, заместитель декана по научной работе факультета информационных технологий и анализа больших данных

Москва

Rasul A. Kochkarov - Cand. Sci. (Econ.), Assoc. Prof. Department of Artificial Intelligence, Deputy Dean for Research at the Faculty of Information Technology and Big Data Analysis

Moscow

rkochkarov@fa.ruФинансовый университет при Правительстве Российской Федерации; ООО «РТК ИТ»Financial University under the Government of the Russian Federation; RTC IT LLCФинансовый университет при Правительстве Российской ФедерацииFinancial University under the Government of the Russian FederationВюрцбургский университет им. Юлиуса МаксимилианаJulius-Maximilians-Universität Würzburg202520112025123243Copyright © Лабинцевa А.И., Кублик Е.И., Перевозчиков Г.П., Кочкаров Р.А., 20252025Лабинцевa А.И., Кублик Е.И., Перевозчиков Г.П., Кочкаров Р.А.Labintsev A.I., Kublik E.I., Perevozchikov G.P., Kоchkarov R.A.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://www.digitarin.ru/jour/article/view/11

В рамках настоящего исследования представлен специализированный датасет Lacmus, разработанный для решения задачи детекции пропавших людей на аэрофотоснимках, полученных с беспилотных летательных аппаратов. Набор данных включает 1552 изображения с более чем 5 тыс. размеченных ограничивающих рамок, зафиксированных в пяти различных локациях, характеризующихся травянистым покровом и редколесьем, в различные сезоны года. Основной целью исследования являлась оптимизация соотношения точности и производительности моделей семейства YOLO v8 на основе представленного датасета. В ходе экспериментальных исследований установлено, что наилучшие показатели достигаются при использовании модели среднего размера с увеличенным входным разрешением изображений без их предварительной обработки и нарезки на снимки меньшего разрешения. Разработанный датасет и полученные результаты исследования предназначены для практического применения в деятельности поисково-спасательных отрядов, что потенциально может способствовать повышению эффективности спасательных операций и спасению человеческих жизней.

This research introduces a specialized Lacmus dataset designed for detecting missing persons in aerial photographs obtained from unmanned aerial vehicles (UAVs). The dataset comprises 1552 images with over 5000 annotated bounding boxes captured at five distinct locations characterized by grassy areas and sparse forests across different seasons. The primary objective of the study was to optimize the accuracy-performance ratio of YOLOv8 models based on the presented dataset. Experimental research has revealed that the best results were achieved using a medium-sized model with increased input image resolution without prior segmentation into smaller resolution images. The developed dataset and research results are intended for practical application in search and rescue operations, which could potentially enhance the efficiency of rescue missions and save human lives.

детекция объектоваэрофотоснимкинейросети YOLOкомпьютерное зрениепоиск людейспасательные операциидатасет Lacmusobject detectionaerial photographsYOLO neural networkscomputer visionperson searchrescue operationslacmus datasetReferencesHerskovits A., Binford T. O. On boundary detection. 1970. URL: http://hdl.handle.net/1721.1/5867Herskovits A., Binford T. O. On boundary detection. 1970. URL: http://hdl.handle.net/1721.1/5867Yakimovsky Y. Boundary and object detection in real world images. Journal of the ACM (JACM). 1976;23(4):599–618. DOI: 10.1145/321978.321981Yakimovsky Y. Boundary and object detection in real world images. Journal of the ACM (JACM). 1976;23(4):599–618. DOI: 10.1145/321978.321981Zou Z., Chen K., Shi Z., Guo Y., Ye J. Object detection in 20 years: A survey. Proceedings of the IEEE. 2023;111(3):257–276. DOI: 10.48550/arXiv.1905.05055Zou Z., Chen K., Shi Z., Guo Y., Ye J. Object detection in 20 years: A survey. Proceedings of the IEEE. 2023;111(3):257–276. DOI: 10.48550/arXiv.1905.05055Carranza-García M., Torres-Mateo J., Lara-Benítez, P. García-Gutiérrez J., On the performance of one-stage and two-stage object detectors in autonomous vehicles using camera data. Remote Sensing. 2020;13(1):89. DOI: 10.3390/rs13010089Carranza-García M., Torres-Mateo J., Lara-Benítez, P. García-Gutiérrez J., On the performance of one-stage and two-stage object detectors in autonomous vehicles using camera data. Remote Sensing. 2020;13(1):89. DOI: 10.3390/rs13010089Shehzadi T., Hashmi K. A., Stricker D., Afzal M. Z. 2D Object Detection with Transformers: A Review. 2023. URL: https://www.researchgate.net/publication/371414252_2D_Object_Detection_with_Transformers_A_ Review. DOI: 10.48550/arXiv.2306.04670Shehzadi T., Hashmi K. A., Stricker D., Afzal M. Z. 2D Object Detection with Transformers: A Review. 2023. URL: https://www.researchgate.net/publication/371414252_2D_Object_Detection_with_Transformers_A_ Review. DOI: 10.48550/arXiv.2306.04670Lin T. Y., Maire M., Belongie S. et al. Microsoft coco: Common objects in context. In Computer Vision — ECCV 2014: 13th European Conference, Zurich, Switzerland, Sept. 6–12, 2014. Proceedings, Springer International Publishing. 2014;3:740–755. URL: https://link.springer.com/book/10.1007/978–3–319–10602–1Lin T. Y., Maire M., Belongie S. et al. Microsoft coco: Common objects in context. In Computer Vision — ECCV 2014: 13th European Conference, Zurich, Switzerland, Sept. 6–12, 2014. Proceedings, Springer International Publishing. 2014;3:740–755. URL: https://link.springer.com/book/10.1007/978–3–319–10602–1Everingham M., Van Gool L., Williams C. K., Winn J. The pascal visual object classes (VOC) challenge. International journal of computer vision. 2010;88:303–338. DOI: 10.1007/s11263–009–0275–4Everingham M., Van Gool L., Williams C. K., Winn J. The pascal visual object classes (VOC) challenge. International journal of computer vision. 2010;88:303–338. DOI: 10.1007/s11263–009–0275–4Zhu P., Wen L., Du D., Bian X. et al. Visdrone-det 2018: The vision meets drone object detection in image challenge results. In Computer Vision — ECCV 2018 Workshops. Lecture Notes in Computer Science. 2019:437–468. DOI: 10.1007/978–3–030–11021–5_27Zhu P., Wen L., Du D., Bian X. et al. Visdrone-det 2018: The vision meets drone object detection in image challenge results. In Computer Vision — ECCV 2018 Workshops. Lecture Notes in Computer Science. 2019:437–468. DOI: 10.1007/978–3–030–11021–5_27Du D., Qi Y., Yu H., et al. The unmanned aerial vehicle benchmark: Object detection and tracking. In Proceedings of the European conference on computer vision (ECCV). 2018:370–386. DOI: 10.48550/arXiv.1804.00518Du D., Qi Y., Yu H., et al. The unmanned aerial vehicle benchmark: Object detection and tracking. In Proceedings of the European conference on computer vision (ECCV). 2018:370–386. DOI: 10.48550/arXiv.1804.00518Varga L. A., Kiefer B., Messmer M., Zell A. Seadronessee: A maritime benchmark for detecting humans in open water. In Proceedings of the IEEE/CVF winter conference on applications of computer vision. 2022:2260–2270. DOI: 10.48550/arXiv.2105.01922Varga L. A., Kiefer B., Messmer M., Zell A. Seadronessee: A maritime benchmark for detecting humans in open water. In Proceedings of the IEEE/CVF winter conference on applications of computer vision. 2022:2260–2270. DOI: 10.48550/arXiv.2105.01922Božić-Štulić D., Marušić Ž. and Gotovac S. Deep learning approach in aerial imagery for supporting land search and rescue missions. International Journal of Computer Vision. 2019;127(9):1–23. DOI: 10.1007/s11263–019–01177–1Božić-Štulić D., Marušić Ž. and Gotovac S. Deep learning approach in aerial imagery for supporting land search and rescue missions. International Journal of Computer Vision. 2019;127(9):1–23. DOI: 10.1007/s11263–019–01177–1Broyles D., Hayner C. R., Leung K. Wisard: A labeled visual and thermal image dataset for wilderness search and rescue. In 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). 2022:9467–9474. DOI: 10.1109/IROS 47612.2022.9981298Broyles D., Hayner C. R., Leung K. Wisard: A labeled visual and thermal image dataset for wilderness search and rescue. In 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). 2022:9467–9474. DOI: 10.1109/IROS 47612.2022.9981298Russell Bernal A. M., Scheirer W., Cleland-Huang J. NOMAD: A Natural, Occluded, Multi-scale Aerial Dataset, for Emergency Response Scenarios. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2024:8584–8595. URL: https://openaccess.thecvf.com/content/WACV2024/papers/Bernal_NOMAD_A_Natural_Occluded_Multi-Scale_Aerial_Dataset_for_Emergency_Response_WACV_2024_paper.pdf. DOI: 10.48550/ arXiv.2309.09518Russell Bernal A. M., Scheirer W., Cleland-Huang J. NOMAD: A Natural, Occluded, Multi-scale Aerial Dataset, for Emergency Response Scenarios. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2024:8584–8595. URL: https://openaccess.thecvf.com/content/WACV2024/papers/Bernal_NOMAD_A_Natural_Occluded_Multi-Scale_Aerial_Dataset_for_Emergency_Response_WACV_2024_paper.pdf. DOI: 10.48550/ arXiv.2309.09518Sambolek S., Ivasic-Kos M., Person Detection and Geolocation Estimation in UAV Aerial Images: An Experimental Approach. In ICPRAM. SN Computer Science. 2025;6(4):785–792. DOI: 10.1007/s42979–025–03869–7Sambolek S., Ivasic-Kos M., Person Detection and Geolocation Estimation in UAV Aerial Images: An Experimental Approach. In ICPRAM. SN Computer Science. 2025;6(4):785–792. DOI: 10.1007/s42979–025–03869–7Akyon F. C., Altinuc S. O., Temizel A. Slicing aided hyper inference and fine-tuning for small object detection. In 2022 IEEE International Conference on Image Processing (ICIP). 2022:966–970. DOI: 10.1109/ICIP46576.2022.9897990Akyon F. C., Altinuc S. O., Temizel A. Slicing aided hyper inference and fine-tuning for small object detection. In 2022 IEEE International Conference on Image Processing (ICIP). 2022:966–970. DOI: 10.1109/ICIP46576.2022.9897990Amjoud A. B., Amrouch M. Object detection using deep learning, CNNs and vision transformers: A review. IEEE Access. 2023;11:35479–35516. DOI: 10.1109/ACCESS.2023.3266093Amjoud A. B., Amrouch M. Object detection using deep learning, CNNs and vision transformers: A review. IEEE Access. 2023;11:35479–35516. DOI: 10.1109/ACCESS.2023.3266093Zhang H., Hao C., Song W., Jiang B., Li B. Adaptive slicing-aided hyper inference for small object detection in highresolution remote sensing images. Remote Sensing. 2023;15(5):1249. DOI: 10.3390/rs15051249Zhang H., Hao C., Song W., Jiang B., Li B. Adaptive slicing-aided hyper inference for small object detection in highresolution remote sensing images. Remote Sensing. 2023;15(5):1249. DOI: 10.3390/rs15051249Labintsev A. I., Dolmatov A. G. Fuzzy measurement of coordinates of small objects in high-resolution images. Soft measurements and calculations. 2022;52(3):36–42. DOI: 10.36871/2618–9976.2022.03.004Labintsev A. I., Dolmatov A. G. Fuzzy measurement of coordinates of small objects in high-resolution images. Soft measurements and calculations. 2022;52(3):36–42. DOI: 10.36871/2618–9976.2022.03.004Redmon J., Divvala S., Girshick R., Farhadi A. You only look once: Unified, real-time object detection. In Proceedings of the Conference: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016. DOI: 10.1109/CVPR.2016.91Redmon J., Divvala S., Girshick R., Farhadi A. You only look once: Unified, real-time object detection. In Proceedings of the Conference: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016. DOI: 10.1109/CVPR.2016.91Hosang J., Benenson R., Schiele B. Learning non-maximum suppression. Proceedings of the IEEE conference on computer vision and pattern recognition. 2017:4507–4515. DOI: 10.1109/CVPR.2017.685Hosang J., Benenson R., Schiele B. Learning non-maximum suppression. Proceedings of the IEEE conference on computer vision and pattern recognition. 2017:4507–4515. DOI: 10.1109/CVPR.2017.685Li X., Wang W., Wu L., Chen S., et al. Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection. Advances in Neural Information Processing Systems. 2020;33:21002–21012. DOI: 10.48550/arXiv.2006.04388Li X., Wang W., Wu L., Chen S., et al. Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection. Advances in Neural Information Processing Systems. 2020;33:21002–21012. DOI: 10.48550/arXiv.2006.04388

The authors declare that there are no conflicts of interest present.