Three-dimensional Damage Visualization of Civil Structures via Gaussian Splatting-enabled Digital Twins

2026-02-20 · Source: cs.CV updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Emerging Technologies & Innovation · Depth: Advanced, quick

Summary

This study introduces a Gaussian Splatting (GS)-enabled digital twin method designed for precise three-dimensional (3D) damage visualization in civil infrastructure. Unlike traditional 2D image-based damage identification or conventional photogrammetric 3D reconstruction, this approach leverages GS for superior scene representation, rendering quality, and efficiency, particularly in featureless areas. Key contributions include using GS-based 3D reconstruction to visualize 2D damage segmentation results while simultaneously reducing segmentation errors. The method also incorporates a multi-scale reconstruction strategy to optimize between efficiency and the capture of fine damage details. Furthermore, it supports dynamic digital twin updates, allowing for the visualization of damage evolution over time. The approach was demonstrated using an open-source synthetic dataset for post-earthquake inspections.

Key takeaway

For civil engineers and infrastructure inspectors needing precise 3D damage visualization, this GS-enabled digital twin method offers a robust solution. You should consider integrating GS-based reconstruction to improve the accuracy of damage segmentation and efficiently track damage evolution over time, moving beyond traditional 2D limitations.

Key insights

Gaussian Splatting enhances 3D damage visualization in civil infrastructure digital twins, improving accuracy and efficiency.

Principles

• Discrete Gaussians outperform continuous implicit models for efficiency.
• Multi-scale reconstruction balances detail and performance.

Method

The method uses GS-based 3D reconstruction to visualize 2D damage segmentation, applies a multi-scale strategy, and enables digital twin updates as damage progresses.

In practice

• Visualize 2D damage segmentation in 3D.
• Reduce segmentation errors via 3D context.
• Update digital twins with evolving damage.

Topics

• Gaussian Splatting
• Digital Twins
• 3D Reconstruction
• Damage Visualization
• Civil Infrastructure

Best for: Computer Vision Engineer, AI Scientist, AI Engineer, Research Scientist, Domain Expert

Related on AIssential

• NG-GS: NeRF-Guided 3D Gaussian Splatting Segmentation — NG-GS improves 3D Gaussian Splatting segmentation by addressing boundary discretization with NeRF-guided optimization.
• MoonSplat: Monocular Online Gaussian Splatting with Sim(3) Global Optimization — MoonSplat enables robust, efficient online 3D reconstruction from monocular video using globally optimized voxelized 3D Gaussian Splatting.
• RPC-GS: Gaussian Splatting with native RPC Rendering for Satellite Imagery — Native RPC integration in Gaussian Splatting significantly improves satellite imagery 3D reconstruction accuracy.
• EnerGS: Energy-Based Gaussian Splatting with Partial Geometric Priors — EnerGS uses a soft energy field from partial geometry to guide 3D Gaussian Splatting, improving reconstruction quality.
• Inside the CUDA Kernel: The GPU Implementation of 3D Gaussian Splatting — Efficient 3DGS rendering relies on parallel processing, optimized data structures, and memory access patterns.
• Towards Next-Generation SLAM: A Survey on 3DGS-SLAM Focusing on Performance, Robustness, and Future Directions — 3D Gaussian Splatting offers a new paradigm for high-fidelity, efficient, and robust SLAM systems.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.