articleProceedings of the AAAI Conference on Artificial IntelligenceApr 11, 2025DIAMOND OA

Segment Any 3D Gaussians

JCJiazhong CenJFJiemin FangCYChen YangLXLingxi XieXZXiaopeng Zhang

Shanghai Jiao Tong University · Huawei Technologies (United Kingdom)

Indexed incrossref

Abstract

This paper presents SAGA (Segment Any 3D GAussians), a highly efficient 3D promptable segmentation method based on 3D Gaussian Splatting (3D-GS). Given 2D visual prompts as input, SAGA can segment the corresponding 3D target represented by 3D Gaussians within 4 ms. This is achieved by attaching a scale-gated affinity feature to each 3D Gaussian to endow it a new property towards multi-granularity segmentation. Specifically, a scale-aware contrastive training strategy is proposed for the scale-gated affinity feature learning. It 1) distills the segmentation capability of the Segment Anything Model (SAM) from 2D masks into the affinity features and 2) employs a soft scale gate mechanism to deal with…

Citation impact

48
total citations
FWCI
30.88
Percentile
100%
References
0
Citations per year

Authors

7

Topics & keywords

Topics
Keywords
  • Mixture model
  • Computer science
  • Artificial intelligence
No related works found for this paper.

Funding