KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#developer-tools #typescript #claude-code #codex #dsh #dsh-plugin #ai-agent #ai-agents #deepseek-harness #agent-skills #macos #cli

LVD_GS-SLAM

★ 36 Open GitHub ↗

No description available.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

// ...existing code...
<p align="center">
  <h1 align="center">LVD-GS: LiDAR-Visual 3D Gaussian Splatting SLAM for Dynamic Scenes</h1>
  <p align="center">
    <a href="https://zwk0901.github.io/LVD-GS2025/"><strong>Project Home — LVD-GS</strong></a>
  </p>
  <p align="center">LVD-GS — hierarchical explicit-implicit representation for robust SLAM in dynamic outdoor scenes</p>

<h3 align="center">Project page / Paper</h3>
  <p align="center">
    Paper (arXiv): https://arxiv.org/abs/2401.12345<br>
    Repo: https://github.com/zwk0901/LVD_GS-SLAM
  </p>
</p>

## :postbox: Updates

- 2025.11.10 In light of my upcoming CSC scholarship interview, I am open-sourcing this initial portion of my work. The rest of the codebase is under active refinement and will be released soon. Thank you for your interest and understanding.Released  initial portion of my work  https://github.com/zwk0901/LVD_GS-SLAM 📝📝📝📝📝📝📝
- 2025.10.25: Released [LVD-GS 中文自动驾驶之心解读](https://mp.weixin.qq.com/s/X9a1fX0HMavOD0uAf732cQ) and [知乎解读：最新SOTA | 东南大学发布LVD-GS：面向智驾，推动3DGS融合动态SLAM](https://www.zhihu.com/question/8989357545/answer/1966801688200447427) introducing LVD-GS! 📝
- 2025.10.23 This paper is released to axiv [LVD-GS (https://arxiv.org/abs/2510.22669].
- 2025.10.22: Released [README.md](README.md)!
- 2025.08.23: This repo is created.
# Getting Started
# Getting Started

## Installation

1. Clone LVD-GS.

```bash
git clone https://github.com/zwk0901/LVD_GS-SLAM.git --recursive
cd LVD_GS-SLAM
```

2. Setup the environment.

```bash
conda env create -f environment.yml
conda activate lvd-gs
```

LVD-GS 测试环境示例：
- Ubuntu 20.04
- PyTorch 2.1.0 / torchvision 0.16.0 / torchaudio 2.1.0（CUDA 11.8）
- GPU: NVIDIA RTX 3090ti

3. Compile submodules for Gaussian splatting

```bash
pip install submodules/simple-knn
pip install submodules/diff-gaussian-rasterization
```

4. Compile the cuda kernels for RoPE (as in CroCo v2 and MASt3R).

```bash
cd croco/models/curope/
python setup.py build_ext --inplace
cd ../../../
```

Our test setup was:

- Ubuntu 20.04: `pytorch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 cudatoolkit=11.8`
- NVIDIA RTX A6000

## Checkpoints

```bash
mkdir -p checkpoints/
wget https://download.europe.naverlabs.com/ComputerVision/MASt3R/MASt3R_ViTLarge_BaseDecoder_512_catmlpdpt_metric.pth -P checkpoints/
```

Please note that you must agree to the MASt3R license when using it.

## Downloading Datasets

#### NuScenes

The processed data for the NuScenes segments can be downloaded via [

Save data under the `datasets/nuscenes` directory.

#### KITTI

The processed sample data for KITTI-07 can be downloaded via

Save data under the `datasets/KITTI` directory.

The full KITTI dataset can be downloaded from [The KITTI Vision Benchmark Suite](https://www.cvlibs.net/datasets/kitti/eval_odometry.php). The specific sequences used in this work are listed in KITTI_sequence_list.txt.

#### Self-collected Dataset

The processed data for the three Self-collected Dataset scenes can be downloaded via
Save data under the `datasets/Self-collected` directory.

## Run

```bash
## NuScenes
CUDA_VISIBLE_DEVICES=0 python slam.py --config "configs/mono/NuScenes/405841.yaml"

## KITTI-01
CUDA_VISIBLE_DEVICES=0 python slam.py --config "configs/mono/KITTI/01.yaml"

##  Self-collected
CUDA_VISIBLE_DEVICES=0 python slam.py --config "configs/mono/Self-collected/2.yaml"
```

## Demo

- If you want to view the real-time interactive SLAM window, please change `Results-use_gui` in `base_config.yaml` to True.

- When running on an Ubuntu system, a GUI window will pop up.

## Run on other dataset

- Please organize your data format and modify the code in `utils/dataset.py`.

- Ground truth depth input interface is still retained in the code, although we didn't use it for SLAM.

# Acknowledgement

- This work is built on [3DGS](https://github.com/graphdeco-inria/gaussian-splatting),  [MonoGS](https://github.com/muskie82/MonoGS), thanks for these great works.

- For more details about Demo, please refer to [MonoGS](https://github.com/muskie82/MonoGS), as we are using its visualization code.

# Citation

If you found our code/work to be useful in your own research, please considering citing the following:

# LVD_GS-SLAM
# motivation
<img width="1863" height="1060" alt="intro" src="https://github.com/user-attachments/assets/525038d8-2a93-4f12-968f-83499bf5c248" />

# pipeline
<img width="2208" height="961" alt="image" src="https://github.com/user-attachments/assets/ac63c40a-f7db-4942-8886-0feb6e48e79c" />

# results

LVD_GS-SLAM

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README