LFX: Fusion-Form-Flexible Light Field Modeling for Dense Semantic Segmentation and Salient Object Detection
LFX: Fusion-Form-Flexible Light Field Modeling for Dense Semantic Segmentation and Salient Object Detection
This research presents LFX, a flexible light field perception framework that supports both sub-aperture image and focal stack fusion settings. By introducing Field-of-Parallax Angular Subspace Modeling (FoP-ASM), the method maps feature differences to observation-specific angular markers without relying on form-specific geometric assumptions or pairwise modeling. A shared manifold subspace with regularization losses ensures semantic consistency across auxiliary observations. Evaluations on three light field datasets show LFX achieves state-of-the-art performance, reducing MAE by 12.12% and 20.59% on DUTLF-V2 and HFUT respectively under sub-aperture and focal stack settings, while achieving the best results on UrbanLF-real. Source code will be released publicly.
The paper introduces LFX, a form-flexible light field perception framework capable of handling both sub-aperture image and focal stack fusion settings via Field-of-Parallax Angular Subspace Modeling, achieving state-of-the-art results on three datasets with reported MAE reductions of up to 20.59% on HFUT under focal stack settings.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
number of LF datasets evaluated
3
View original evidence
n and the central image to an observation-specific angular marker. This avoids pairwise modeling among auxiliary observations and form-specific geometric assumptions. Meanwhile, a shared manifold subspace and the associated regularization losses constrain the angular residuals within a common feature space, promoting semantic consistency across auxiliary observations. Extensive evaluations across three LF datasets sh
Open source S4
- code: Reported pending Open source S5
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ow that LFX achieves state-of-the-art results across distinct LF forms. Under the S.Aperture images and FS settings, respectively, LFX reduces MAE by 12.12%/20.59% on DUTLF-V2 and by 6.15%/7.69% on HFUT, while achieving the best performance on UrbanLF-real. The source code will be made publicly available at https://github.com/FeiT-FeiTeng/LFX.
Open source S5
n and the central image to an observation-specific angular marker. This avoids pairwise modeling among auxiliary observations and form-specific geometric assumptions. Meanwhile, a shared manifold subspace and the associated regularization losses constrain the angular residuals within a common feature space, promoting semantic consistency across auxiliary observations. Extensive evaluations across three LF datasets sh
Open source S4
LFX: Fusion-Form-Flexible Light Field Modeling for Dense Semantic Segmentation and Salient Object Detection
Open source S1
Source：arXiv Robotics — research abstracts · arxiv.org