EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Open Set Vehicle Detection for UAV-Based Images Using an Out-of-Distribution Detector

Fei Zhao, Wenzhong Lou, Yi Sun, Zihao Zhang, Wenlong Ma, Chenglong Li

Drones · 30.06.2023 · 7 (7) · 434

DOI: 10.3390/drones7070434 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Drones, online erschienen am 30.06.2023. Verfasst von Fei Zhao und 5 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Natural Science Foundation of Chongqing.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Vehicle target detection is a key technology for reconnaissance unmanned aerial vehicles (UAVs). However, in order to obtain a larger reconnaissance field of view, this type of UAV generally flies at a higher altitude, resulting in a relatively small proportion of vehicle targets in its imaging images. Moreover, due to the unique nature of the mission, previously unseen vehicle types are prone to appearing in the surveillance area. Additionally, it is challenging for large-scale detectors based on deep learning to achieve real-time performance on UAV computing equipment. To address these problems, we propose a vehicle object detector specifically designed for UAVs in this paper. We have made modifications to the backbone of Faster R-CNN based on the target and scene characteristics. We have improved the positioning accuracy of small-scale imaging targets by adjusting the size and ratio of anchors. Furthermore, we have introduced a postprocessing method for out-of-distribution detection, enabling the designed detector to detect and distinguish untrained vehicle types. Additionally, to tackle the scarcity of reconnaissance images, we have constructed two datasets using modeling and image rendering techniques. We have evaluated our method on these constructed datasets. The proposed method achieves a 96% mean Average Precision at IoU threshold 0.5 (mAP50) on trained objects and a 71% mAP50 on untrained objects. Equivalent flight experiments demonstrate that our model, trained on synthetic data, can achieve satisfactory detection performance and computational efficiency in practical applications.

Förderung

  • Natural Science Foundation of Chongqing

Zitieren

APA

Zhao, F., Lou, W., Sun, Y., Zhang, Z., Ma, W., & Li, C. (2023). Open Set Vehicle Detection for UAV-Based Images Using an Out-of-Distribution Detector. Drones, 7(7), 434. https://doi.org/10.3390/drones7070434

Vancouver

Zhao F, Lou W, Sun Y, Zhang Z, Ma W, Li C. Open Set Vehicle Detection for UAV-Based Images Using an Out-of-Distribution Detector. Drones. 2023;7(7):434. doi:10.3390/drones7070434

BibTeX
@article{Zhao2023,
  author = {Zhao, Fei and Lou, Wenzhong and Sun, Yi and Zhang, Zihao and Ma, Wenlong and Li, Chenglong},
  title = {Open Set Vehicle Detection for UAV-Based Images Using an Out-of-Distribution Detector},
  journal = {Drones},
  year = {2023},
  volume = {7},
  number = {7},
  pages = {434},
  doi = {10.3390/drones7070434}
}
Kostenlos starten →