arXiv cs.AI· Dung Do Tien, Kien Nguyen Do Trung, Kiet Dang Vu, Trung Thai Tran, Duc Dung Nguyen·· 11 小时前AI 评分18
OccSplat：面向驾驶场景重建的遮挡感知高斯致密化
Quo Vadis, Driving Scene Reconstruction? Occlusion-Aware Gaussian Densification
AI 导读
OccSplat 是一个前馈框架，核心贡献是遮挡感知高斯致密化，直接针对缺失的场景几何而非缺失像素，在 3D 表示中补全场景。它先通过 Gaussian Occlusion Mapping 识别缺乏高斯支撑的遮挡区域，再由 Gaussian Occlusion Completion 生成新图元并整合进原始表示，保持无需逐场景优化的前馈推理。
正文 · 原文
Abstract:While recent feed-forward 3D Gaussian Splatting methods enable efficient driving-scene reconstruction from surround-view images, their reliance on a single observation fundamentally hinders novel-view synthesis. Regions occluded in the input can become visible from novel viewpoints, leaving the reconstructed Gaussians under-constrained and incomplete. Moving beyond image-space inpainting for rendered views, we introduce OccSplat, a feed-forward framework whose core contribution is Occlusion-Aware Gaussian Densification, explicitly targeting missing scene geometry rather than missing pixels. By predicting new Gaussian primitives for unsupported regions, OccSplat completes the scene directly in the 3D representation. Specifically, OccSplat first identifies occluded regions that lack Gaussian support through Gaussian Occlusion Mapping. Gaussian Occlusion Completion then generates new primitives for these regions. The predicted primitives are integrated into the original representation to improve novel-view synthesis while retaining feed-forward inference without per-scene optimization. Experiments on the nuScenes benchmark demonstrate that OccSplat improves on the compared methods and yields particularly large gains in occluded regions, leading to more accurate and visually coherent driving-scene reconstruction.
| Comments: | Accepted at the Workshop on Computer Vision for Developing Countries (CV4DC), ACCV 2026 (Proceedings Track) |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.06273 [cs.CV] |
| (or arXiv:2611.06273v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.06273 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Kiet Dang Vu [view email]
[v1]
Wed, 4 Nov 2026 10:17:04 UTC (6,940 KB)
来源：arXiv cs.AI · arxiv.org