gaussian-splatting-engineering

Links: GitHub Link

Authors: *EvangelionA

Motivation

This repository aims to bridge the gap between academic research and industrial deployment of 3D Gaussian Splatting (3DGS) technology. While most existing papers focus on theoretical innovations, this project emphasizes:

  • Engineering-oriented Implementation - Providing practical, scalable 3DGS solutions with a focus on performance and stability in real-world application scenarios.
  • Steady Performance Optimization - Systematically improving and optimizing 3DGS to enhance operational efficiency across various hardware configurations and environments.
  • Systematic Documentation of Practical Workflows - Comprehensive documentation of the complete pipeline from data acquisition and camera pose estimation to 3DGS training and application.

Our project is built upon the original 3DGS.

Bilibili: gaussian-splatting-engineering[Introduction](项目介绍)

Dataset

Dataset Model PSNR SSIM
KITTI 3DGS_base 28.50 ----
ABS-GS 28.55 ----
RAIN-GS ---- ----
smith_hall_outdoor 3DGS_base 24.85 ----
ABS-GS 24.64 ----
RAIN-GS ---- ----

Qualitative Results

WIP

Workflow

Bilibili: gaussian-splatting-engineering[Introduction](PIPLINE梳理)

Gaussian Splatting Engineering Pipeline

Stage_0. Data Acquisition

Stage_1. Camera Pose Estimation

Stage_2. 3DGS Training

Stage_3. Export & Deployment

Bibtex

@inproceedings{li2025scenesplat,
  title={gaussian-splatting-engineering},
  author={EvangelionA},
  year={2025}
}

Acknowledgement

This project is built upon 3DGS and This Project-webpage is built upon SceneSplat We thank all the authors for their great repos.

LICENSE

*This website is licensed under the CC BY-SA 4.0 License. The image comparison slider is from the Dics repository.