EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Systematic Benchmark of Object Detectors for Small-Object Detection in Side-Scan Sonar Imagery Under Small-Sample Conditions

Lei Liu, Houpu Li, Hua Ouyang, Song Li, Guojun Zhai

Journal of Marine Science and Engineering · 05.11.2026 · 14 (21) · 2090

DOI: 10.3390/jmse14212090 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Marine Science and Engineering, online erschienen am 05.11.2026. Verfasst von Lei Liu und 4 weiteren Autor:innen. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Small object detection in side-scan sonar imagery is hindered by heavy speckle noise, limited target texture, and severe data scarcity. Existing studies mostly transplant improvements designed for natural images to sonar scenarios, and their effectiveness under small-sample conditions lacks unified verification. This study presents a systematic benchmark of lightweight object detectors on a publicly available side-scan sonar small-object dataset, SSS-MLOD. Under a unified protocol, five YOLO variants are compared under from-scratch training and COCO-pretrained transfer learning, and 25 structural, loss-based, and data-side modifications are evaluated. The results show that differences among lightweight baselines are marginal (mAP50 of 0.732 to 0.753), transfer learning on only 243 training images yields the largest single gain (mAP50 from 0.742 to 0.856 for YOLOv13n), and none of the 25 modifications was observed to improve the baseline under from-scratch training on this dataset. Increasing the input resolution improves mAP50:95 by 0.044 to 0.049 for YOLOv5n and YOLOv8n but degrades the performance of YOLOv11n, indicating that the benefit of resolution is architecture-dependent; test-time augmentation with a tuned confidence threshold raises recall by 0.035 under a precision constraint of 0.85. The best configuration, YOLOv13n with transfer learning, achieves an mAP50 of 0.856 and a recall of 0.757. These findings provide an evidence-based reference for detector selection and configuration in resource constrained applications such as autonomous underwater vehicles.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Liu, L., Li, H., Ouyang, H., Li, S., & Zhai, G. (2026). A Systematic Benchmark of Object Detectors for Small-Object Detection in Side-Scan Sonar Imagery Under Small-Sample Conditions. Journal of Marine Science and Engineering, 14(21), 2090. https://doi.org/10.3390/jmse14212090

Vancouver

Liu L, Li H, Ouyang H, Li S, Zhai G. A Systematic Benchmark of Object Detectors for Small-Object Detection in Side-Scan Sonar Imagery Under Small-Sample Conditions. Journal of Marine Science and Engineering. 2026;14(21):2090. doi:10.3390/jmse14212090

BibTeX
@article{Liu2026,
  author = {Liu, Lei and Li, Houpu and Ouyang, Hua and Li, Song and Zhai, Guojun},
  title = {A Systematic Benchmark of Object Detectors for Small-Object Detection in Side-Scan Sonar Imagery Under Small-Sample Conditions},
  journal = {Journal of Marine Science and Engineering},
  year = {2026},
  volume = {14},
  number = {21},
  pages = {2090},
  doi = {10.3390/jmse14212090}
}
Kostenlos starten →