EUVIMEDEUVIMED entdecken
Crossref

Beyond Wide-Angle Images: Structure-to-Detail Video Portrait Correction via Unsupervised Spatiotemporal Adaptation

Wenbo Nie, Lang Nie, Chunyu Lin, Jingwen Chen, Ke Xing, Jiyuan Wang, Kang Liao

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (10) · 8151-8159

DOI: 10.1609/aaai.v40i10.37762 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Wenbo Nie und 6 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Wide-angle cameras, despite their popularity for content creation, suffer from distortion-induced facial stretching—especially at the edge of the lens—which degrades visual appeal. To address this issue, we propose a structure-to-detail portrait correction model named ImagePC. It integrates the long-range awareness of the transformer and multi-step denoising of diffusion models into a unified framework, achieving global structural robustness and local detail refinement. Besides, considering the high cost of obtaining video labels, we then repurpose ImagePC for unlabeled wide-angle videos (termed VideoPC), by spatiotemporal diffusion adaption with spatial consistency and temporal smoothness constraints. For the former, we encourage the denoised image to approximate pseudo labels following the wide-angle distortion distribution pattern, while for the latter, we derive rectification trajectories with backward optical flows and smooth them. Compared with ImagePC, VideoPC maintains high-quality facial corrections in space and mitigates the potential temporal shakes sequentially in blind scenarios. Finally, to establish an evaluation benchmark and train the framework, we establish a video portrait dataset with a large diversity in the number of people, lighting conditions, and background. Experiments demonstrate that the proposed methods outperform existing solutions quantitatively and qualitatively, contributing to high-fidelity wide-angle videos with stable and natural portraits.

Zitieren

APA

Nie, W., Nie, L., Lin, C., Chen, J., Xing, K., Wang, J., & Liao, K. (2026). Beyond Wide-Angle Images: Structure-to-Detail Video Portrait Correction via Unsupervised Spatiotemporal Adaptation. Proceedings of the AAAI Conference on Artificial Intelligence, 40(10), 8151-8159. https://doi.org/10.1609/aaai.v40i10.37762

Vancouver

Nie W, Nie L, Lin C, Chen J, Xing K, Wang J, et al. Beyond Wide-Angle Images: Structure-to-Detail Video Portrait Correction via Unsupervised Spatiotemporal Adaptation. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(10):8151-8159. doi:10.1609/aaai.v40i10.37762

BibTeX
@article{Nie2026,
  author = {Nie, Wenbo and Nie, Lang and Lin, Chunyu and Chen, Jingwen and Xing, Ke and Wang, Jiyuan and Liao, Kang},
  title = {Beyond Wide-Angle Images: Structure-to-Detail Video Portrait Correction via Unsupervised Spatiotemporal Adaptation},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {10},
  pages = {8151--8159},
  doi = {10.1609/aaai.v40i10.37762}
}
Kostenlos starten →