Pith. sign in

REVIEW 13 cited by

LangSplat: 3D Language Gaussian Splatting

Not yet reviewed by Pith; the record is open.

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 2312.16084 v2 pith:NUB2N463 submitted 2023-12-26 cs.CV

classification cs.CV
keywords languagelangsplatfieldfeaturesclipefficientembeddingsexisting
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
abstract

Humans live in a 3D world and commonly use natural language to interact with a 3D scene. Modeling a 3D language field to support open-ended language queries in 3D has gained increasing attention recently. This paper introduces LangSplat, which constructs a 3D language field that enables precise and efficient open-vocabulary querying within 3D spaces. Unlike existing methods that ground CLIP language embeddings in a NeRF model, LangSplat advances the field by utilizing a collection of 3D Gaussians, each encoding language features distilled from CLIP, to represent the language field. By employing a tile-based splatting technique for rendering language features, we circumvent the costly rendering process inherent in NeRF. Instead of directly learning CLIP embeddings, LangSplat first trains a scene-wise language autoencoder and then learns language features on the scene-specific latent space, thereby alleviating substantial memory demands imposed by explicit modeling. Existing methods struggle with imprecise and vague 3D language fields, which fail to discern clear boundaries between objects. We delve into this issue and propose to learn hierarchical semantics using SAM, thereby eliminating the need for extensively querying the language field across various scales and the regularization of DINO features. Extensive experimental results show that LangSplat significantly outperforms the previous state-of-the-art method LERF by a large margin. Notably, LangSplat is extremely efficient, achieving a 199 $\times$ speedup compared to LERF at the resolution of 1440 $\times$ 1080. We strongly recommend readers to check out our video results at https://langsplat.github.io/

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 13 Pith papers

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

  1. Splat and Replace: 3D Reconstruction with Repetitive Elements

    cs.GR 2025-06 conditional novelty 7.0 of 10

    Repetitive objects in 3D scenes are registered into a shared Gaussian representation that propagates well-observed geometry and appearance to poorly observed instances, improving rendered novel views.

  2. LabelGS: Label-Aware 3D Gaussian Splatting for 3D Scene Segmentation

    cs.CV 2025-08 conditional novelty 6.0 of 10

    LabelGS assigns 2D video-tracking labels to the most-contributing 3D Gaussians, with depth-based occlusion masking and a projection filter, reporting better mIoU/PSNR than Feature-3DGS with roughly 22x faster training.

  3. Predict-Optimize-Distill: A Self-Improving Cycle for 4D Object Understanding

    cs.CV 2025-04 conditional novelty 6.0 of 10

    POD interleaves prediction, inverse-rendering optimization, and self-distillation to reconstruct articulated object part poses from monocular video, outperforming an optimization-only baseline and improving with video...

  4. Latent Radiance Fields with 3D-aware 2D Representations

    cs.CV 2025-02 conditional novelty 6.0 of 10

    A three-stage pipeline makes VAE latent codes 3D-consistent and builds a latent radiance field, improving photorealistic novel-view synthesis in latent space.

  5. SparseLGS: Sparse View Language Embedded Gaussian Splatting

    cs.CV 2024-12 conditional novelty 6.0 of 10

    SparseLGS builds open-vocabulary 3D semantic fields from 3-4 pose-free images, using learning-based pose/point-cloud estimation and a three-step multi-view semantic alignment.

  6. SparseGrasp: Robotic Grasping via 3D Semantic Gaussian Splatting from Sparse Multi-View RGB Images

    cs.RO 2024-12 conditional novelty 6.0 of 10

    A language-guided robotic grasping system that reconstructs a 3D semantic scene from three RGB views and updates moved objects in about 200 ms, reporting higher grasp success than F3RM and LERF-TOGO.

  7. Bootstraping Clustering of Gaussians for View-consistent 3D Scene Understanding

    cs.CV 2024-11 conditional novelty 6.0 of 10

    FreeGS bootstraps view-consistent semantics and instance indices in 3D Gaussian Splatting without needing 2D masks.

  8. Gradient-Weighted Feature Back-Projection: A Fast Alternative to Feature Distillation in 3D Gaussian Splatting

    cs.CV 2024-11 conditional novelty 6.0 of 10

    A gradient-weighted back-projection of 2D features onto 3D Gaussians yields fast 2D/3D segmentation and identity encoding at quality comparable to trained feature-field methods.

  9. Leveraging 2D Priors and SDF Guidance for Dynamic Urban Scene Rendering

    cs.CV 2025-10 conditional novelty 5.0 of 10

    UGSDF achieves state-of-the-art novel-view rendering of dynamic urban objects without LiDAR or 3D motion annotations by jointly optimizing SDFs and 3D Gaussians under 2D depth and point-tracking priors.

  10. A Neural Representation Framework with LLM-Driven Spatial Reasoning for Open-Vocabulary 3D Visual Grounding

    cs.CV 2025-07 conditional novelty 5.0 of 10

    SpatialReasoner adds LLM-based query decomposition and a visual-properties-enhanced hierarchical feature field to 3D language fields, improving instance localization for spatial-relation queries on self-constructed be...

  11. GaussianProperty: Integrating Physical Properties to 3D Gaussians with LMMs

    cs.RO 2024-12 conditional novelty 5.0 of 10

    A training-free pipeline uses SAM segmentation and GPT-4V material recognition, then votes across views to attach density, elasticity, and friction values to 3D Gaussians for simulation and grasping.

  12. From Coarse to Fine: Learnable Discrete Wavelet Transforms for Efficient 3D Gaussian Splatting

    cs.CV 2025-06 conditional novelty 4.0 of 10

    AutoOpti3DGS uses learnable discrete wavelet transforms on input images to train 3DGS from coarse to fine, reducing peak Gaussian counts by about 18 to 23 percent with modest quality trade-offs.

  13. OpenMaskDINO3D : Reasoning 3D Segmentation via Large Language Model

    cs.CV 2025-06 reject novelty 3.0 of 10

    OpenMaskDINO3D reports state-of-the-art 3D reasoning segmentation with a LISA-style SEG token and object identifiers, but uses Mask3D pseudo-labels as ground truth and lacks released code.

Pith tools