Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SwiftGS: Episodic Priors for Immediate Satellite Surface Recovery

Rong Fu, Jiekai Wu, Haiyun Wei, Xiaowen Ma, Shiyin Lin, Kangan Qian, Chuang Liu, Jianyuan Ni, Simon James Fong

Abstract

Rapid, large-scale 3D reconstruction from multi-date satellite imagery is vital for environmental monitoring, urban planning, and disaster response, yet remains difficult due to illumination changes, sensor heterogeneity, and the cost of per-scene optimization. We introduce SwiftGS, a meta-learned system that reconstructs 3D surfaces in a single forward pass by predicting geometry-radiation-decoupled Gaussian primitives together with a lightweight SDF, replacing expensive per-scene fitting with episodic training that captures transferable priors. The model couples a differentiable physics graph for projection, illumination, and sensor response with spatial gating that blends sparse Gaussian detail and global SDF structure, and incorporates semantic-geometric fusion, conditional lightweight task heads, and multi-view supervision from a frozen geometric teacher under an uncertainty-aware multi-task loss. At inference, SwiftGS operates zero-shot with optional compact calibration and achieves accurate DSM reconstruction and view-consistent rendering at significantly reduced computational cost, with ablations highlighting the benefits of the hybrid representation, physics-aware rendering, and episodic meta-training.

SwiftGS: Episodic Priors for Immediate Satellite Surface Recovery

Abstract

Rapid, large-scale 3D reconstruction from multi-date satellite imagery is vital for environmental monitoring, urban planning, and disaster response, yet remains difficult due to illumination changes, sensor heterogeneity, and the cost of per-scene optimization. We introduce SwiftGS, a meta-learned system that reconstructs 3D surfaces in a single forward pass by predicting geometry-radiation-decoupled Gaussian primitives together with a lightweight SDF, replacing expensive per-scene fitting with episodic training that captures transferable priors. The model couples a differentiable physics graph for projection, illumination, and sensor response with spatial gating that blends sparse Gaussian detail and global SDF structure, and incorporates semantic-geometric fusion, conditional lightweight task heads, and multi-view supervision from a frozen geometric teacher under an uncertainty-aware multi-task loss. At inference, SwiftGS operates zero-shot with optional compact calibration and achieves accurate DSM reconstruction and view-consistent rendering at significantly reduced computational cost, with ablations highlighting the benefits of the hybrid representation, physics-aware rendering, and episodic meta-training.
Paper Structure (52 sections, 5 theorems, 54 equations, 6 figures, 15 tables, 1 algorithm)

This paper contains 52 sections, 5 theorems, 54 equations, 6 figures, 15 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Neural rendering for satellite photogrammetry
  4. Geometric calibration and physical illumination modeling
  5. Meta-learning and transferable priors for reconstruction
  6. Geometric supervision and scalable system design
  7. Positioning relative to contemporary methods
  8. Methodology
  9. Problem statement and notation
  10. Core architecture overview
  11. Unified differentiable physics graph
  12. Representation: Gaussian primitive and hybrid mixture
  13. Encoder, scene latent and hybrid decoder
  14. Gating and conditional activation
  15. Lightweight task-specific heads and conditional computation
  16. ...and 37 more sections

Key Result

Proposition A.1

Suppose the true radiance--density field $\mathcal{W}^\star$ admits the decomposition where $\lambda_{\mathrm{g}}^\star$ and $\lambda_{\mathrm{s}}^\star$ are measurable functions satisfying $\lambda_{\mathrm{g}}^\star(x)+\lambda_{\mathrm{s}}^\star(x)=1$ for every $x$. Let the model class contain predictors $\mathcal{W}_{\Phi}$ indexed by parameters $\Phi$, and denote by $\lambda_g(\c and training

Figures (6)

  • Figure 1: Overview of the SwiftGS architecture for efficient, zero-shot satellite surface reconstruction. The pipeline begins with Multi-View Encoding, where per-view features and a global scene latent are extracted. A Hybrid Decoder produces a compact Gaussian set $\Gamma$, an implicit SDF $S_{\psi}$, and spatial gates that blend sparse and dense components. Lightweight Task-Specific Heads optionally refine geometry or appearance. A Differentiable Physics Graph renders elevation and albedo using camera geometry and illumination, supported by distilled geometric cues during training. An episodic meta-training scheme learns shared parameters $\Phi$ with a small per-scene calibration vector $\theta$. At inference, SwiftGS runs zero-shot and outputs a DSM, consistent renderings, and a compact Gaussian memory for incremental reconstruction.
  • Figure 2: Shadow and lighting consistency: input, predicted shadow, rendered image, albedo, and ground truth.
  • Figure 3: Training curves showing stable convergence of query loss, DSM error, and shadow loss.
  • Figure 4: Qualitative comparison across scenes. Each row shows input images, predicted DSM and renderings from SwiftGS and EO-NeRF, along with ground truth. MAE and LPIPS errors are annotated.
  • Figure 5: Representative failure cases of SwiftGS. Top row: dense high-rise urban canyon exhibiting height hallucination due to severe occlusion. Middle row: water body with systematic elevation overestimation caused by specular reflection violating diffuse reflectance assumptions. Bottom row: extreme shadow coverage ($>$60%) showing incomplete shadow removal and associated height artifacts. For each case, columns show input image, predicted DSM, error map (red indicates overestimation, blue underestimation), and ground truth DSM.
  • ...and 1 more figures

Theorems & Definitions (7)

  • Proposition A.1
  • Proposition A.2
  • Lemma A.3
  • Proposition A.4
  • proof
  • Proposition A.5
  • proof