R4DGS: Referring Segmentation in 4D Gaussian Splatting

Chen Bangpu, Author 2, Author 3, Author 4, Author 5

4D Gaussian Splattingmultimodaldynamic scenesnatural language processing

2610.30326

Builder Relevance

80%

2h ago

Abstract

This paper introduces R4DGS, a method for grounding complex natural-language referring in dynamic 4D scenes using a unified framework called ReferGaussian.

Reality Card

Core Claim

R4DGS achieves improved temporal target selection and competitive spatiotemporal grounding without the need for scene-specific semantic training.

Method / Result

Demonstrated improved performance on the R4D-Bench-QA benchmark.

Limitations

The method may still face challenges in generalizing to highly complex or novel dynamic scenes not represented in the training data.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.

← Back to all papers