Enregistré dans:
Détails bibliographiques
Auteurs principaux: Lv, Chengxin, Li, Yihui, Yang, Hongyu, Wang, YunHong
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2603.22852
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866917359401304064
author Lv, Chengxin
Li, Yihui
Yang, Hongyu
Wang, YunHong
author_facet Lv, Chengxin
Li, Yihui
Yang, Hongyu
Wang, YunHong
contents 3D semantic occupancy prediction is crucial for autonomous driving. While multi-modal fusion improves accuracy over vision-only methods, it typically relies on computationally expensive dense voxel or BEV tensors. We present Gau-Occ, a multi-modal framework that bypasses dense volumetric processing by modeling the scene as a compact collection of semantic 3D Gaussians. To ensure geometric completeness, we propose a LiDAR Completion Diffuser (LCD) that recovers missing structures from sparse LiDAR to initialize robust Gaussian anchors. Furthermore, we introduce Gaussian Anchor Fusion (GAF), which efficiently integrates multi-view image semantics via geometry-aligned 2D sampling and cross-modal alignment. By refining these compact Gaussian descriptors, Gau-Occ captures both spatial consistency and semantic discriminability. Extensive experiments across challenging benchmarks demonstrate that Gau-Occ achieves state-of-the-art performance with significant computational efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2603_22852
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Gau-Occ: Geometry-Completed Gaussians for Multi-Modal 3D Occupancy Prediction
Lv, Chengxin
Li, Yihui
Yang, Hongyu
Wang, YunHong
Computer Vision and Pattern Recognition
3D semantic occupancy prediction is crucial for autonomous driving. While multi-modal fusion improves accuracy over vision-only methods, it typically relies on computationally expensive dense voxel or BEV tensors. We present Gau-Occ, a multi-modal framework that bypasses dense volumetric processing by modeling the scene as a compact collection of semantic 3D Gaussians. To ensure geometric completeness, we propose a LiDAR Completion Diffuser (LCD) that recovers missing structures from sparse LiDAR to initialize robust Gaussian anchors. Furthermore, we introduce Gaussian Anchor Fusion (GAF), which efficiently integrates multi-view image semantics via geometry-aligned 2D sampling and cross-modal alignment. By refining these compact Gaussian descriptors, Gau-Occ captures both spatial consistency and semantic discriminability. Extensive experiments across challenging benchmarks demonstrate that Gau-Occ achieves state-of-the-art performance with significant computational efficiency.
title Gau-Occ: Geometry-Completed Gaussians for Multi-Modal 3D Occupancy Prediction
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.22852