sapiСистемный анализ и прикладная информатика«System analysis and applied information science»2309-49232414-0481Belarusian National Technical University10.21122/2309-4923-2025-3-4-10sapi-757Research ArticleСистемный анализSystem analysisМодель обнаружения объектов на изображениях дистанционного зондирования земли с использованием динамического рецептивного поля и Snake-сверткиRemote sensing image target detection model integrating dynamic receptive field and snake convolutionВуC.WuX.

Ву Сяньи – аспирант механико-математического факультета.

г. Минск, 220030

Wu Xiangyi – Postgraduate student of the Faculty of Mechanics and Mathematics.

Minsk, 220030 

tigerv5872@gmail.comАбламейкоС. В.AblameykoS. B.

Абламейко Сергей Владимирович – Академик, доктор технических наук, профессор, Лауреат Государственной премии.

г. Минск, 220012

 

 

Ablameyko Sergey Vladimirovich – Academician, Doctor of Science (Engineering), Professor. Laureate of the State Prize.

 Minsk, 220012

ablameyko@bsu.byБелорусский государственный университетБеларусьBelarusian State UniversityBelarusОбъединенный институт проблем информатики Национальной академии наук БеларусиБеларусьThe United Institute of Informatics Problems of the National Academy of Sciences of BelarusBelarus20251610202503410Copyright © Ву C., Абламейко С.В., 20252025Ву C., Абламейко С.В.Wu X., Ablameyko S.B.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://sapi.bntu.by/jour/article/view/757

Для решения задачи обнаружения объектов на изображениях дистанционного зондирования Земли (ДЗЗ) в данной работе предлагается усовершенствованный метод на базе YOLOv11n. Предложена улучшенная архитектура YOLOv11n, интегрирующая модуль динамического рецептивного поля (RFAConv) и модуль адаптивного моделирования деформаций Snake (DySnakeConv). Этот подход улучшает процесс выявления низкоуровневых признаков и оптимизирует адаптивное выделение границ объектов, повышая точность обнаружения объектов. Эксперименты на наборе данных RSOD показали, что улучшенная модель достигает средней точности (mAP) 96.9 % при IoU = 0.50 (mAP50) и 65.5 % в диапазоне IoU 0.50–0.95 (mAP50-95). Результаты превосходят показатели YOLOv8n, YOLOv10n и других конкурентных моделей по ключевым метрикам (точности и полноте). Важно отметить, что модель сохраняет сопоставимую эффективность на наборе NWPU VHR-10. Предложенная модель является эффективным решением для обнаружения малых объектов и геометрически сложных целей на изображениях ДЗЗ высокого разрешения.

To address the challenges of a high missed detection rate for small targets and strong interference from complex backgrounds in remote sensing image target detection, the paper proposes an improved YOLOv11n based method. We introduce an enhanced YOLOv11n model incorporating a dynamic receptive field module (RFAConv) and a snake deformation modeling module (DySnakeConv). This approach strengthens shallow feature extraction capabilities and refines adaptive fitting of target boundaries, thereby improving detection accuracy. Experimental results demonstrate that on the RSOD dataset, the improved model achieves mean average precision (mAP) scores of 96.9 % at IoU = 0.50 (mAP50) and 65.5 % over IoU thresholds from 0.50 to 0.95 (mAP5095). These results surpass those of YOLOv8n, YOLOv10n, and other comparative models in key metrics such as precision and recall. Importantly, the model maintains comparable performance on the NWPU VHR-10 dataset. The proposed model presents an efficient solution for detecting small and geometrically sensitive targets in high-resolution remote sensing images.

изображения дистанционного зондирования Землиобнаружение объектовYOLOv11RFAConvDySnakeConvRemote Sensing ImageObject DetectionYOLOv11RFAConvDySnakeConvReferencesLi AB, Guo H, Qi C, et al. Dense object detection in remote sensing images under complex background. Computer Engineering and Applications. 2023;59(8):247–253.Li AB, Guo H, Qi C, et al. Dense object detection in remote sensing images under complex background. Computer Engineering and Applications. 2023;59(8):247–253.Yan J, Hu X, Zhang K, Shi T, Zhu G, Zhang Y. Multi-level feature fusion based dim small ground target detection in remote sensing images. Chinese Journal of Scientific Instrument. 2022;4(3):221–229. https://library.imaging.org/jist/articles/67/1/010505Yan J, Hu X, Zhang K, Shi T, Zhu G, Zhang Y. Multi-level feature fusion based dim small ground target detection in remote sensing images. Chinese Journal of Scientific Instrument. 2022;4(3):221–229. https://library.imaging.org/jist/articles/67/1/010505Ren S, He K, Girshick R, Sun J. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2017;39(6),1137–1149. DOI: 10.1109/TPAMI.2016.2577031Ren S, He K, Girshick R, Sun J. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2017;39(6),1137–1149. DOI: 10.1109/TPAMI.2016.2577031Jang Y, Gunes H, Patras I. Registration-free Face-SSD: Single shot analysis of smiles, facial attributes, and affect in the wild. Computer Vision and Image Understanding. 2019;182:17–29. DOI: 10.1016/j.cviu.2019.01.006Jang Y, Gunes H, Patras I. Registration-free Face-SSD: Single shot analysis of smiles, facial attributes, and affect in the wild. Computer Vision and Image Understanding. 2019;182:17–29. DOI: 10.1016/j.cviu.2019.01.006Lin T-Y, Goyal P, Girshick R, He K, Dollár P. Focal Loss for Dense Object Detection. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2020;42(2):318–327. DOI: 10.1109/TPAMI.2018.2858826Lin T-Y, Goyal P, Girshick R, He K, Dollár P. Focal Loss for Dense Object Detection. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2020;42(2):318–327. DOI: 10.1109/TPAMI.2018.2858826Redmon J, Divvala S, Girshick R, Farhadi A. (2016). You Only Look Once: Unified, Real-Time Object Detection. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA; 2016. Рp. 779-788. DOI: 10.1109/CVPR.2016.91Redmon J, Divvala S, Girshick R, Farhadi A. (2016). You Only Look Once: Unified, Real-Time Object Detection. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA; 2016. Рp. 779-788. DOI: 10.1109/CVPR.2016.91Betti A, Tucci M. YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery. Sensors. 2023;23(4):1865. DOI: 10.3390/s23041865Betti A, Tucci M. YOLO-S: A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery. Sensors. 2023;23(4):1865. DOI: 10.3390/s23041865Wu X, Ablameyko SV. Efficient detection of building in remote sensing images using an improved YOLOv10 network. Informatika [Informatics]. 2025;22(2):33–47. DOI: 10.37661/1816-0301-2025-22-2-33-47Wu X, Ablameyko SV. Efficient detection of building in remote sensing images using an improved YOLOv10 network. Informatika [Informatics]. 2025;22(2):33–47. DOI: 10.37661/1816-0301-2025-22-2-33-47Han W, Kuerban A, Yang Y, Huang Z, Liu B, Gao J. (2022). Multi-Vision Network for Accurate and Real-Time Small Object Detection in Optical Remote Sensing Images. IEEE Geoscience and Remote Sensing Letters. 2022;19:1–5. DOI: 10.1109/LGRS.2020.3044422Han W, Kuerban A, Yang Y, Huang Z, Liu B, Gao J. (2022). Multi-Vision Network for Accurate and Real-Time Small Object Detection in Optical Remote Sensing Images. IEEE Geoscience and Remote Sensing Letters. 2022;19:1–5. DOI: 10.1109/LGRS.2020.3044422Nie H, Pang H, Ma M, Zheng R. A Lightweight Remote Sensing Small Target Image Detection Algorithm Based on Improved YOLOv8. Sensors (Basel). 2024;24(9):2952. DOI: 10.3390/s24092952Nie H, Pang H, Ma M, Zheng R. A Lightweight Remote Sensing Small Target Image Detection Algorithm Based on Improved YOLOv8. Sensors (Basel). 2024;24(9):2952. DOI: 10.3390/s24092952Wu Z, Geiger A., Rozner J, Kreiss E, Lu H, Icard T, Potts C, & Goodman ND. Training trajectories of language models across scales. 2023. arXiv preprint arXiv:2304.03198. https://arxiv.org/pdf/2304.03198Wu Z, Geiger A., Rozner J, Kreiss E, Lu H, Icard T, Potts C, & Goodman ND. Training trajectories of language models across scales. 2023. arXiv preprint arXiv:2304.03198. https://arxiv.org/pdf/2304.03198Mirchandani S, Xia F, Florence P, Ichter B, Driess D, Arenas MG, et al. Large Language Models as General Pattern Machines. Proceedings of the 7th Conference on Robot Learning (CoRL). Atlanta, USA; 2023. pp. 2498-2518. Available at: https://general-pattern-machines.github.io (accessed 02.07.2025). DOI: 10.48550/arXiv.2307.04721Mirchandani S, Xia F, Florence P, Ichter B, Driess D, Arenas MG, et al. Large Language Models as General Pattern Machines. Proceedings of the 7th Conference on Robot Learning (CoRL). Atlanta, USA; 2023. pp. 2498-2518. Available at: https://general-pattern-machines.github.io (accessed 02.07.2025). DOI: 10.48550/arXiv.2307.04721Long Y, Gong Y, Xiao Z, Liu Q. Accurate Object Localization in Remote Sensing Images Based on Convolutional Neural Networks. IEEE Transactions on Geoscience and Remote Sensing. 2017;55(5):2486-2498. DOI: 10.1109/TGRS.2016.2645610Long Y, Gong Y, Xiao Z, Liu Q. Accurate Object Localization in Remote Sensing Images Based on Convolutional Neural Networks. IEEE Transactions on Geoscience and Remote Sensing. 2017;55(5):2486-2498. DOI: 10.1109/TGRS.2016.2645610Cheng G, Han J, Zhou P, Guo L. Multi-class geospatial object detection and geographic image classification based on collection of part detectors. ISPRS Journal of Photogrammetry and Remote Sensing. 2014;98:119–132. DOI: 10.1016/j.isprsjprs.2014.10.002Cheng G, Han J, Zhou P, Guo L. Multi-class geospatial object detection and geographic image classification based on collection of part detectors. ISPRS Journal of Photogrammetry and Remote Sensing. 2014;98:119–132. DOI: 10.1016/j.isprsjprs.2014.10.002

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