EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Detection of Low‐Altitude Unmanned Vehicle and Flying Bird by Improved YOLOv7

Weiyu Xie, Qiang Zhang, Tanwuguo Chen

International Journal of Aerospace Engineering · 2026-01 · 2026 (1)

DOI: 10.1155/ijae/1802637 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Aerospace Engineering, online erschienen am 21.07.2026. Verfasst von Weiyu Xie und 2 weiteren Autor:innen. Die Arbeit nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: China Scholarship Council.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

With the rapid development of low‐altitude economy and unmanned aerial vehicle (UAV) technologies, reliable UAV surveillance has become increasingly important for aviation safety and low‐altitude airspace management. However, accurate recognition of UAVs in optical imagery remains challenging due to small target sizes, motion blur, complex backgrounds, and the visual similarity between UAVs and flying birds. To address these challenges, this paper proposes an improved YOLOv7‐based framework for low‐altitude UAV and bird recognition. The proposed framework adopts MobileNetV3 as a lightweight backbone to reduce computational complexity while maintaining efficient feature extraction capability. To improve small‐target detection performance, a modified BiFPN and SPD‐Conv based small‐object enhancement module are introduced to enhance multiscale feature fusion and preserve fine‐grained spatial information. Furthermore, a trajectory‐aware recognition strategy based on a simplified Global Tracking Transformer (GTR) is incorporated to exploit temporal motion characteristics and improve UAV–bird discrimination in dynamic scenes. In addition, a dedicated UAV–Bird dataset containing realistic low‐altitude surveillance scenarios is constructed for training and evaluation. Extensive experiments are conducted on multiple public datasets, including MAV‐VID, Anti‐UAV RGB, and an open source Drone vs. Birds Dataset from Kaggle, as well as the proposed dataset. The results demonstrate that the proposed framework achieves competitive detection performance while maintaining real‐time inference capability. On the self‐built dataset, the proposed method improves mAP@0.5 from 0.927 to 0.933 and F1 score from 0.903 to 0.914 compared with the original YOLOv7 while increasing the inference speed from 33.6 to 55.6 FPS. These results indicate that the proposed framework effectively balances detection accuracy and computational efficiency, making it suitable for practical low‐altitude surveillance applications.

Förderung

  • China Scholarship Council

Zitieren

APA

Xie, W., Zhang, Q., & Chen, T. (2026). Detection of Low‐Altitude Unmanned Vehicle and Flying Bird by Improved YOLOv7. International Journal of Aerospace Engineering, 2026(1). https://doi.org/10.1155/ijae/1802637

Vancouver

Xie W, Zhang Q, Chen T. Detection of Low‐Altitude Unmanned Vehicle and Flying Bird by Improved YOLOv7. International Journal of Aerospace Engineering. 2026;2026(1). doi:10.1155/ijae/1802637

BibTeX
@article{Xie2026,
  author = {Xie, Weiyu and Zhang, Qiang and Chen, Tanwuguo},
  title = {Detection of Low‐Altitude Unmanned Vehicle and Flying Bird by Improved YOLOv7},
  journal = {International Journal of Aerospace Engineering},
  year = {2026},
  volume = {2026},
  number = {1},
  doi = {10.1155/ijae/1802637}
}
Kostenlos starten →