This repository provides the official implementation of the paper “ GS2E: Gaussian Splatting is an Effective Data Generator for Event Stream Generation ”, which introduces a novel pipeline that leverages 3D Gaussian Splatting to manipulate camera trajectories, render photorealistic image sequences, and simulate dense, high-fidelity event streams for advancing event-based vision research.
We showcase a selection of paired RGB-Event samples generated by GS2E, along with visualizations of the simulated event streams as videos.
- We have released a portion of the dataset on Hugging Face.
- 🚧 The full codebase is being cleaned and documented.
- Once accepted, we will release all code and datasets.
GS2E offers a fully implemented pipeline for constructing custom event datasets from rendered image sequences.
With support for trajectory editing, event simulation, and visualization, users can easily adapt the framework to generate high-quality event streams tailored to their own scenes.
Follow the steps below to set up the environment and install dependencies.
-
Clone the repository:
git clone git@github.com:PKU-YuanGroup/GS2E.git cd GS2E -
Install the necessary dependencies:
This project is based on 3D Gaussian Splatting, so please refer to its installation instructions.
Some parts of this repository rely on additional packages that were used during the exploratory phase. These components are not essential for reproducing the main results and can be safely ignored during typical usage.
We would like to thank all contributors and collaborators who supported this project. Special thanks to the authors of 3D Gaussian Splatting for their foundational work, which inspired and enabled this research.
If you find our work useful in your research, please consider citing:
@misc{li2025gs2egaussiansplattingeffective,
title={GS2E: Gaussian Splatting is an Effective Data Generator for Event Stream Generation},
author={Yuchen Li and Chaoran Feng and Zhenyu Tang and Kaiyuan Deng and Wangbo Yu and Yonghong Tian and Li Yuan},
year={2025},
eprint={2505.15287},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2505.15287},
}