EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

SaLon3R: Structure-Aware Long-Term Feedforward 3D Reconstruction from Unposed Images

Jiaxin Guo, Tongfan Guan, Wenzhen Dong, Wenzhao Zheng, Wenting Wang, Yue Wang, Yeung Yam, Yun-Hui Liu

International Journal of Computer Vision · 2026-08 · 134 (8)

DOI: 10.1007/s11263-026-02965-2 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Computer Vision, online erschienen am 05.08.2026. Verfasst von Jiaxin Guo und 7 weiteren Autor:innen. Die Arbeit nennt 85 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Shenzhen Portion of Shenzhen-Hong Kong Science and Technology Innovation Cooperation Zone, HK RGC AoE, InnoHK initiative of the Innovation and Technology Commission of the Hong Kong Special Administrative Region Government via the Hong Kong Centre for Logistics….

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Recent advances in 3D Gaussian Splatting (3DGS) have enabled feed-forward, on-the-fly reconstruction of sequential input views. However, existing methods often predict per-pixel Gaussians and combine Gaussians from all views as the scene representation, leading to substantial redundancies and geometric inconsistencies in long-duration video sequences. To address this, we propose SaLon3R , a novel framework for S tructure- a ware, Lon g-term 3 DGS R econstruction. Our method eliminates redundancy by introducing compact anchor primitives as a replacement for per-pixel Gaussians. These primitives are derived through a differentiable, saliency-aware Gaussian quantization process designed to preserve fidelity while ensuring a compact representation. Specifically, a foundational 3D reconstruction model is employed to predict a saliency map encoding regional geometric complexity. Guided by this saliency map, we compress redundant Gaussian primitives into compact anchors by prioritizing high-complexity regions. Furthermore, we introduce a 3D Point Transformer to overcome geometric inconsistencies caused by long-term accumulative errors. It refines attributes and saliency of the anchor primitives leveraging the learned spatial structural priors in 3D space. Without known camera parameters or test-time optimization, our approach effectively prunes the redundant 3DGS and resolves artifacts in a single feed-forward pass. Experiments on multiple datasets demonstrate our approach outperform state-of-the-arts on both novel view synthesis and depth estimation, while exhibiting superior efficiency ( $$>\!$$ > 10 ), robustness, and generalization ability for long-term feed-forward Gaussian reconstruction.

Förderung

  • Shenzhen Portion of Shenzhen-Hong Kong Science and Technology Innovation Cooperation Zone
  • HK RGC AoE
  • InnoHK initiative of the Innovation and Technology Commission of the Hong Kong Special Administrative Region Government via the Hong Kong Centre for Logistics…
  • Multi-Scale Medical Robotics Centre
  • CUHK T Stone Robotics Institute

Zitieren

APA

Guo, J., Guan, T., Dong, W., Zheng, W., Wang, W., Wang, Y., Yam, Y., & Liu, Y. (2026). SaLon3R: Structure-Aware Long-Term Feedforward 3D Reconstruction from Unposed Images. International Journal of Computer Vision, 134(8). https://doi.org/10.1007/s11263-026-02965-2

Vancouver

Guo J, Guan T, Dong W, Zheng W, Wang W, Wang Y, et al. SaLon3R: Structure-Aware Long-Term Feedforward 3D Reconstruction from Unposed Images. International Journal of Computer Vision. 2026;134(8). doi:10.1007/s11263-026-02965-2

BibTeX
@article{Guo2026,
  author = {Guo, Jiaxin and Guan, Tongfan and Dong, Wenzhen and Zheng, Wenzhao and Wang, Wenting and Wang, Yue and Yam, Yeung and Liu, Yun-Hui},
  title = {SaLon3R: Structure-Aware Long-Term Feedforward 3D Reconstruction from Unposed Images},
  journal = {International Journal of Computer Vision},
  year = {2026},
  volume = {134},
  number = {8},
  doi = {10.1007/s11263-026-02965-2}
}
Kostenlos starten →