EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Generative AI and Simulation-Based Data Augmentation for Enhanced Object Detection in Low-Data Forestry Environments

Krzysztof Wołk, Ramana Reddy Avula, Aleksi Narkilahti, Marek S. Tatara, Jacek Niklewski, Oleg Żero

Forests · 27.02.2026 · 17 (3) · 302

DOI: 10.3390/f17030302 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Forests, online erschienen am 27.02.2026. Verfasst von Krzysztof Wołk und 5 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Chips Joint Undertaking, National Centre for Research and Development of Poland.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Detecting rare ground-level obstacles (e.g., large boulders) in dense boreal forests from low-altitude UAV RGB imagery is challenging due to limited annotated data, strong background clutter, and expensive field labeling. This paper evaluates two complementary synthetic-data augmentation pipelines for low-data forestry object detection: segmentation-guided diffusion inpainting, where SegFormer-derived forest-floor masks constrain Stable Diffusion inpainting to plausible insertion regions, and simulator-based generation in Unreal Engine 5 with controlled domain randomization and automatic annotations. We conduct a ten-fold cross-validation study on a real UAV dataset of 64 images and report both accuracy and stability across folds. Compared to real-only training (mean mAP50 ≈ 0.579; mAP50-95 ≈ 0.350), inpainting improves mean performance (mAP50 ≈ 0.647; mAP50-95 ≈ 0.435) while substantially reducing cross-fold variance and lifting the worst-case fold from 0.301 to 0.619 in mAP50. Simulator augmentation yields slightly lower mean accuracy (mAP50 ≈ 0.546; mAP50-95 ≈ 0.344) but markedly improves robustness by mitigating collapse on difficult splits (minimum mAP50 0.496 vs. 0.301). These results indicate that carefully curated generative augmentation can reduce failure risk and improve generalization in extremely data-limited forestry detection settings without additional field data collection.

Förderung

  • Chips Joint Undertaking
  • National Centre for Research and Development of Poland

Zitieren

APA

Wołk, K., Avula, R. R., Narkilahti, A., Tatara, M. S., Niklewski, J., & Żero, O. (2026). Generative AI and Simulation-Based Data Augmentation for Enhanced Object Detection in Low-Data Forestry Environments. Forests, 17(3), 302. https://doi.org/10.3390/f17030302

Vancouver

Wołk K, Avula RR, Narkilahti A, Tatara MS, Niklewski J, Żero O. Generative AI and Simulation-Based Data Augmentation for Enhanced Object Detection in Low-Data Forestry Environments. Forests. 2026;17(3):302. doi:10.3390/f17030302

BibTeX
@article{Wok2026,
  author = {Wołk, Krzysztof and Avula, Ramana Reddy and Narkilahti, Aleksi and Tatara, Marek S. and Niklewski, Jacek and Żero, Oleg},
  title = {Generative AI and Simulation-Based Data Augmentation for Enhanced Object Detection in Low-Data Forestry Environments},
  journal = {Forests},
  year = {2026},
  volume = {17},
  number = {3},
  pages = {302},
  doi = {10.3390/f17030302}
}
Kostenlos starten →