Saved in:
Bibliographic Details
Main Authors: Li, Yihui, Lv, Chengxin, Tang, Zichen, Yang, Hongyu, Huang, Di
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.00697
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915825576837120
author Li, Yihui
Lv, Chengxin
Tang, Zichen
Yang, Hongyu
Huang, Di
author_facet Li, Yihui
Lv, Chengxin
Tang, Zichen
Yang, Hongyu
Huang, Di
contents We present TokenSplat, a feed-forward framework for joint 3D Gaussian reconstruction and camera pose estimation from unposed multi-view images. At its core, TokenSplat introduces a Token-aligned Gaussian Prediction module that aligns semantically corresponding information across views directly in the feature space. Guided by coarse token positions and fusion confidence, it aggregates multi-scale contextual features to enable long-range cross-view reasoning and reduce redundancy from overlapping Gaussians. To further enhance pose robustness and disentangle viewpoint cues from scene semantics, TokenSplat employs learnable camera tokens and an Asymmetric Dual-Flow Decoder (ADF-Decoder) that enforces directionally constrained communication between camera and image tokens. This maintains clean factorization within a feed-forward architecture, enabling coherent reconstruction and stable pose estimation without iterative refinement. Extensive experiments demonstrate that TokenSplat achieves higher reconstruction fidelity and novel-view synthesis quality in pose-free settings, and significantly improves pose estimation accuracy compared to prior pose-free methods. Project page: https://kidleyh.github.io/tokensplat/.
format Preprint
id arxiv_https___arxiv_org_abs_2603_00697
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TokenSplat: Token-aligned 3D Gaussian Splatting for Feed-forward Pose-free Reconstruction
Li, Yihui
Lv, Chengxin
Tang, Zichen
Yang, Hongyu
Huang, Di
Computer Vision and Pattern Recognition
We present TokenSplat, a feed-forward framework for joint 3D Gaussian reconstruction and camera pose estimation from unposed multi-view images. At its core, TokenSplat introduces a Token-aligned Gaussian Prediction module that aligns semantically corresponding information across views directly in the feature space. Guided by coarse token positions and fusion confidence, it aggregates multi-scale contextual features to enable long-range cross-view reasoning and reduce redundancy from overlapping Gaussians. To further enhance pose robustness and disentangle viewpoint cues from scene semantics, TokenSplat employs learnable camera tokens and an Asymmetric Dual-Flow Decoder (ADF-Decoder) that enforces directionally constrained communication between camera and image tokens. This maintains clean factorization within a feed-forward architecture, enabling coherent reconstruction and stable pose estimation without iterative refinement. Extensive experiments demonstrate that TokenSplat achieves higher reconstruction fidelity and novel-view synthesis quality in pose-free settings, and significantly improves pose estimation accuracy compared to prior pose-free methods. Project page: https://kidleyh.github.io/tokensplat/.
title TokenSplat: Token-aligned 3D Gaussian Splatting for Feed-forward Pose-free Reconstruction
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.00697