SplatPose: Geometry-Aware 6-DoF Pose Estimation from Single RGB Image via 3D Gaussian Splatting

2025-03-07

AI Summary

Existing 6-DoF pose estimation methods from a single RGB image suffer from strong initialization dependence, high rotational ambiguity, and—critically—reliance on depth sensors or multi-view inputs, leading to high deployment costs. Method: This paper proposes a pure-RGB, high-precision pose estimation framework that integrates differentiable rendering of 3D Gaussian Splatting (3DGS) with a dual-branch neural network. We introduce a novel geometric-domain attention mechanism to decouple position and pose alignment, and design a coarse-to-fine optimization pipeline jointly correcting 2D feature misalignment and sparse ray-depth errors. Contribution/Results: Our method achieves state-of-the-art performance on three standard benchmarks under the single-RGB setting, matching the accuracy of advanced depth- or multi-view–based approaches while significantly reducing hardware requirements and deployment complexity.

Technology Category

• Computer Vision: Biometrics, Face, Gesture & Pose
• Intelligent Robots: State Estimation
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

6-DoF pose estimation is a fundamental task in computer vision with wide-ranging applications in augmented reality and robotics. Existing single RGB-based methods often compromise accuracy due to their reliance on initial pose estimates and susceptibility to rotational ambiguity, while approaches requiring depth sensors or multi-view setups incur significant deployment costs. To address these limitations, we introduce SplatPose, a novel framework that synergizes 3D Gaussian Splatting (3DGS) with a dual-branch neural architecture to achieve high-precision pose estimation using only a single RGB image. Central to our approach is the Dual-Attention Ray Scoring Network (DARS-Net), which innovatively decouples positional and angular alignment through geometry-domain attention mechanisms, explicitly modeling directional dependencies to mitigate rotational ambiguity. Additionally, a coarse-to-fine optimization pipeline progressively refines pose estimates by aligning dense 2D features between query images and 3DGS-synthesized views, effectively correcting feature misalignment and depth errors from sparse ray sampling. Experiments on three benchmark datasets demonstrate that SplatPose achieves state-of-the-art 6-DoF pose estimation accuracy in single RGB settings, rivaling approaches that depend on depth or multi-view images.

Problem

• Improves 6-DoF pose estimation from single RGB images
• Reduces reliance on depth sensors or multi-view setups
• Mitigates rotational ambiguity and initial pose dependency

Innovation

• Uses 3D Gaussian Splatting for pose estimation
• Implements Dual-Attention Ray Scoring Network
• Employs coarse-to-fine optimization pipeline