Pith. sign in

REVIEW 10 cited by

GaussianFormer: Scene as Gaussians for Vision-Based 3D Semantic Occupancy Prediction

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2405.17429 v1 pith:3S3ZLTZW submitted 2024-05-27 cs.CV cs.AI

Yuanhui Huang , Wenzhao Zheng , Yunpeng Zhang , Jie Zhou , Jiwen Lu This is my paper
classification cs.CVcs.AI
keywords gaussiansoccupancysemanticgaussianformerscenemethodsonlyposition
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

3D semantic occupancy prediction aims to obtain 3D fine-grained geometry and semantics of the surrounding scene and is an important task for the robustness of vision-centric autonomous driving. Most existing methods employ dense grids such as voxels as scene representations, which ignore the sparsity of occupancy and the diversity of object scales and thus lead to unbalanced allocation of resources. To address this, we propose an object-centric representation to describe 3D scenes with sparse 3D semantic Gaussians where each Gaussian represents a flexible region of interest and its semantic features. We aggregate information from images through the attention mechanism and iteratively refine the properties of 3D Gaussians including position, covariance, and semantics. We then propose an efficient Gaussian-to-voxel splatting method to generate 3D occupancy predictions, which only aggregates the neighboring Gaussians for a certain position. We conduct extensive experiments on the widely adopted nuScenes and KITTI-360 datasets. Experimental results demonstrate that GaussianFormer achieves comparable performance with state-of-the-art methods with only 17.8% - 24.8% of their memory consumption. Code is available at: https://github.com/huang-yh/GaussianFormer.

Discussion (0). Sign in to comment.

Forward citations

Cited by 10 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. VISA: VLM-Guided Instance Semantic Auditing for 3D Occupancy World Models

    cs.CV 2026-06 unverdicted novelty 7.0 of 10

    VISA improves closed-set 3D occupancy mIoU on nuScenes by using VLM instance audits as reliability-weighted semantic supervisors during training of existing world models.

  2. FPSGen: Flexible Point Cloud Scene Generation with BEV-Supported Transport Flows

    cs.CV 2026-07 conditional novelty 6.5 of 10

    A two-stage flow-matching model that seeds point-cloud generation from a generated BEV density map, using teacher-estimated point pairings to keep transport paths straight, achieves SOTA JSD/IoU on SemanticKITTI compl...

  3. SparseOcc++: Geometry-Aware Sparse Latent Representation for Semantic Occupancy Prediction

    cs.CV 2026-07 accept novelty 6.5 of 10

    SparseOcc++ decouples geometry completion (via orthogonal SCF regression on sparse anchors) from semantics, improving IoU 2.3 points and running 3.9 imes faster than SparseOcc on nuScenes while 5.9 imes faster than Oc...

  4. GaussianSeed: Hierarchical Gaussian Seeding for High-Resolution 3D Occupancy Prediction

    cs.CV 2026-07 conditional novelty 6.0 of 10

    A hierarchical Gaussian occupancy representation with regression-based seeding predicts high-resolution 3D occupancy at lower latency than prior sparse baselines, validated on nuScenes and a new 0.1m campus dataset.

  5. RayOcc: Occlusion-Aware Ray Occupancy Estimation via Gaussian Mixture Intensity

    cs.CV 2026-07 conditional novelty 6.0 of 10

    RayOcc models each camera ray as a non-normalized Gaussian mixture with Poisson-based occupancy probabilities, allowing multiple depth hypotheses per ray and improving Gaussian-initialized 3D occupancy prediction on nuScenes.

  6. Self-Supervised Sparse Sensor Fusion for Long Range Perception

    cs.CV 2025-08 conditional novelty 6.0 of 10

    LRS4Fusion fuses cameras and LiDAR in a fully sparse voxel representation with self-supervised temporal pre-training, achieving 52.61 mAP for detection out to 250 meters.

  7. VoxelSplat: Dynamic Gaussian Splatting as an Effective Loss for Occupancy and Flow Prediction

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A training-only Gaussian splatting loss, which renders predicted 3D semantics and motion into 2D camera views, improves semantic occupancy and scene flow prediction across several camera-based models.

  8. OcRFDet: Object-Centric Radiance Fields for Multi-View 3D Object Detection in Autonomous Driving

    cs.CV 2025-06 conditional novelty 5.0 of 10

    Adding object-centric radiance-field rendering and height-aware opacity attention to the DualBEV detector improves camera-only 3D object detection on nuScenes by up to 2.0 mAP points.

  9. QuadricFormer: Scene as Superquadrics for 3D Semantic Occupancy Prediction

    cs.CV 2025-06 conditional novelty 5.0 of 10

    QuadricFormer represents 3D scenes as a probabilistic mixture of superquadrics, improving accuracy and efficiency over Gaussian-based occupancy prediction on nuScenes.

  10. R3GS: Gaussian Splatting for Robust Reconstruction and Relocalization in Unconstrained Image Collections

    cs.CV 2025-05 conditional novelty 5.0 of 10

    R3GS integrates appearance-conditioned hash features, a fine-tuned human-detector visibility map, and a fixed sky sphere into 3D Gaussian Splatting to improve novel view synthesis and relocalization on Phototourism scenes.

Pith tools