B$^3$-Seg: Camera-Free, Training-Free 3DGS Segmentation via Analytic EIG and Beta-Bernoulli Bayesian Updates

Hiromichi Kamata; Samuel Arthur Munro; Fuminori Homma

cs.CVFeb 19, 2026

B $^3$ -Seg: Camera-Free, Training-Free 3DGS Segmentation via Analytic EIG and Beta-Bernoulli Bayesian Updates

Authors: Hiromichi Kamata, Samuel Arthur Munro, Fuminori Homma

Organizations: Sony Group Corporation, Tokyo, Japan · Pixomondo, London, United Kingdom

Abstract

Interactive 3D Gaussian Splatting (3DGS) segmentation is essential for real-time editing of pre-reconstructed assets in film and game production. However, existing methods rely on predefined camera viewpoints, ground-truth labels, or costly retraining, making them impractical for low-latency use. We propose B $^3$ -Seg (Beta-Bernoulli Bayesian Segmentation for 3DGS), a fast and theoretically grounded method for open-vocabulary 3DGS segmentation under camera-free and training-free conditions. Our approach reformulates segmentation as sequential Beta-Bernoulli Bayesian updates and actively selects the next view via analytic Expected Information Gain (EIG). This Bayesian formulation guarantees the adaptive monotonicity and submodularity of EIG, which produces a greedy $(1{-}1/e)$ approximation to the optimal view sampling policy. Experiments on multiple datasets show that B $^3$ -Seg achieves competitive results to high-cost supervised methods while operating end-to-end segmentation within a few seconds. The results demonstrate that B $^3$ -Seg enables practical, interactive 3DGS segmentation with provable information efficiency.

Figures & tables

Figure 1: B 3 -Seg actively selects the next view via Expected Information Gain and updates 3D labels by Beta–Bernoulli Bayesian updates. It runs in a few seconds and requires neither predefined views nor ground-truth semantic labels.

Figure 2: Overview of B 3 -Seg . (1) Sample $N_{\mathrm{cand}}$ candidate views on a sphere centered at the estimated object center $\mathbf{c}_{\mathrm{obj}}$ . (2) Render each candidate to compute $\mathrm{EIG}(v)$ by Eq. ( 11 ), and pick the best view $v^{\star}$ (red). On $v^{\star}$ , obtain masks using Grounded SAM2 and CLIP reranking. (3) From the mask, compute $(e_{i,1},e_{i,0})$ by Eq. ( 5 ) and update Beta parameters. We iterate (1)–(3) process 20 steps. The pipeline enables camera-free, training-free, open-vocabulary 3DGS segmentation in a few seconds.

Figure 3: Information Gain vs. Expected Information Gain (ours). (a) IG calculation updates the Beta posterior using SAM2 segmentation masks (Eq. ( 9 )). (b) Our EIG approximates the posterior update from the prior Beta distribution, avoiding SAM2 inference and enabling efficient viewpoint evaluation (Eq. ( 11 )).

Figure 4: Qualitative comparison on text-guided 3D segmentation. We compare our method ( B 3 -Seg ) with prior 3DGS segmentation approaches. Our method produces cleaner and more complete object masks, especially in cluttered scenes.

Figure 5: Candidate-view EIG on LERF-Mask (Teatime) with the prompt “stuffed bear” . Each panel shows a candidate rendering; the bottom-right inset is the current confidence map (posterior mean).

Method	Accuracy (mIoU / mBIoU)				views	time	steps
Method	figurines	ramen	teatime	mean	views	time	steps
Assumes reconstruction views and/or labels (not directly comparable)
LERF [ 13 ]	33.5 / 30.6	28.3 / 14.7	49.7 / 42.6	37.2 / 29.3	GT	45 min	30k
SA3D [ 3 ]	24.9 / 23.8	7.4 / 7.0	42.5 / 39.2	24.9 / 23.3	GT	35 min	30k
LangSplat [ 19 ]	52.8 / 50.5	50.4 / 44.7	69.5 / 65.6	57.6 / 53.6	GT	19 min	30k
Gaussian Grouping [ 29 ]	69.7 / 67.9	77.0 / 68.8	71.7 / 66.1	72.8 / 67.6	GT	37 min	30k

Table 1: LERF-Mask (accuracy, assumptions, and latency). Top: Methods that require reconstruction views/labels (=not directly comparable). Bottom: Sampling-based, training-free approach with our 20 views/updates runtime (few seconds). † Uniform-Sphere : Candidate viewpoints sampled uniformly on a sphere. ‡ Recon-Cam : Candidate viewpoints randomly sampled from reconstruction cameras.

Assumes reconstruction views/labels (not comparable)
Method	mIoU (%)
Method	Bed	Bench	Sofa	Lawn
LangSplat [ 19 ]	92.5	94.2	90.0	96.1
Feature 3DGS [ 32 ]	83.5	90.7	86.9	93.4
LEGaussians [ 24 ]	84.9	91.1	87.8	92.5
Gaussian Grouping [ 29 ]	83.0	91.5	87.3	90.6

Table 2: 3D - OVS (mIoU). Top: assumes views/labels ( not comparable ). Bottom: camera-free & training-free.

CLIP re-rank	SAM2 mask-input	Accuracy		$\Delta$ vs. Base
CLIP re-rank	SAM2 mask-input	mIoU	mBIoU	mIoU	mBIoU
–	–	74.9	70.2	–	–
✓	–	81.5	76.6	+6.6	+6.4
✓	✓	84.5	81.0	+9.6	+10.8

Table 3: Ablation study on LERF-Mask. Both CLIP re-ranking and SAM2 mask-input improve segmentation performance.

Figure 6: Predicted EIG closely matches information gain on the LERF-Mask, with a strong correlation ( $r=0.964$ ).

Figure 7: Posterior entropy vs. iteration for three view-selection strategies on the LERF-Mask Teatime scene. EIG-based selection consistently achieves the largest per-step entropy drop.

	Mask Inference	View Selection	Beta Update	Other	Total
Time (s)	9.761	2.111	0.119	0.117	12.108

Table 4: Runtime breakdown of B 3 -Seg . End-to-end inference completes in 12 seconds for 20 actively selected views.

$N_{\mathrm{cand}}$	5	10	20	30
mIoU (%)	76.2	83.1	84.5	84.6
time (s)	10.3	11.6	12.1	14.4
$T$ (iterations)	5	10	20	30
mIoU (%)	79.8	82.7	84.5	84.8
time (s)	3.5	6.6	12.1	19.4

Table 5: Sensitivity to

N_{\mathrm{cand}}

and

T

on LERF-Mask. Accuracy saturates around

N_{\mathrm{cand}}{=}20

and

T{=}20

while runtime increases almost linearly.

$\mathbf{c}_{\mathrm{obj}}$ shift	0%	20%	50%	70%	100%
mIoU (%)	84.5	83.6	82.9	81.7	80.7
$\Delta$	–	-0.9	-1.6	-2.8	-3.8

Table 6: Sensitivity to the initial condition on LERF-Mask. We shift the initial object center

\mathbf{c}_{\mathrm{obj}}

along a random 3D direction by a percentage of

r_{\mathrm{obj}}

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 8: Example of assumed application of B 3 -Seg. A user specifies an object in an interactive 3DGS editor through text, clicking, or dragging. Our method then performs fast and stable 3D segmentation from the current viewpoint.

Figure 9: Effect of CLIP re-ranking in the LERF-Mask Teatime scene. Although GroundingDINO assigns a higher score to the wrong bounding box (green), CLIP correctly assigns a higher similarity score to the region corresponding to the true object described by the prompt “cookies on a plate” (orange).

Figure 10: Qualitative results of B 3 -Seg on LERF-Mask

Figure 11: Qualitative results of B 3 -Seg on 3D-OVS

Figure 12: Qualitative results of other 3D scenes

3DGS Segmentation

See all 15 papers

Online Segment 3D Gaussians via Launching Virtual Drones
Jul 2, 2026Liwei Liao, Rongjie Wang, Ronggang Wang3D Gaussian Splatting 3DGS Segmentation
Guangdong Provincial Key Laboratory of Ultra High Definition Immersive Media Technology, Shenzhen Graduate School, Peking University, Shenzhen, China · Pengcheng Laboratory, Shenzhen, China · Peking University, Beijing, China
Rendering-Aware Bayesian 3D Gaussian Splatting with Native Uncertainty and Adaptive Complexity Control
Jul 6, 2026Gaoxiang Jia, Vikram Appia, Junzhou Huang +1Active View Selection Active Perception
Advanced Micro Devices, Inc. · Dept. of Computer Science and Engineering University of Texas at Arlington · Dept. of Mathematics & Division of Data Science University of Texas at Arlington
Seed2GS: Camera-Free, Training-Free Object Extraction from 3D Gaussian Scenes via a Single Reference-View Grounding
Aug 12, 2026Zongjian Ding, Yudong Gao, Jiale Liu +73D Gaussian Splatting Open-Vocabulary 3D Segmentation
University of Chinese Academy of Sciences · Institute of Information Engineering, Chinese Academy of Sciences · 4The Hong Kong University of Science and Technology +2

Jul 2, 2026J/K move · Enter open · S save

Online Segment 3D Gaussians via Launching Virtual Drones

Liwei Liao, Rongjie Wang, Ronggang Wang

Guangdong Provincial Key Laboratory of Ultra High Definition Immersive Media Technology, Shenzhen Graduate School, Peking University, Shenzhen, China · Pengcheng Laboratory, Shenzhen, China · Peking University, Beijing, China

3D Gaussian Splatting 3DGS Segmentation

Open paper AskPDF arXiv

Interactive segmentation of 3D Gaussians offers a compelling opportunity for real-time manipulation of 3D scenes, thanks to the real-time rendering capability of 3D Gaussian Splatting (3DGS). However, existing methods require a time-consuming per-scene setup - typically tens of seconds or even minutes - before interactive segmentation can begin on a raw 3DGS scene. This setup involves multi-view mask preparation, mask lifting, and feature distillation, creating a major bottleneck for online applications. To address this limitation, we aim to completely eliminate the setup stage for interactive 3DGS segmentation while keeping the segmentation time practical (under 1 second). In this work, we present SAGO (Segment Any Gaussians Online), a novel setup-free framework for interactive 3DGS segmentation. By introducing virtual drones, our method reframes the 3D segmentation problem as an online Next-Best-View (NBV) planning task formulated within a Markov process. Extensive experiments demonstrate that SAGO can extract clean 3D assets directly from 3D Gaussians with sub-second latency, thereby enabling a broad range of downstream applications such as object manipulation and scene editing. Moreover, our method achieves over a 50x speedup compared to the previous setup-free 3DGS segmentation frameworks.

Abstract

3DGS Segmentation

Explore similar work

Online Segment 3D Gaussians via Launching Virtual Drones

Rendering-Aware Bayesian 3D Gaussian Splatting with Native Uncertainty and Adaptive Complexity Control

Seed2GS: Camera-Free, Training-Free Object Extraction from 3D Gaussian Scenes via a Single Reference-View Grounding