Yansong Xu · Junlin Li · Wei Zhang · Siyu Chen · Shengyong Zhang · Yuquan Leng^∗ · Weijia Zhou^∗

FGS-SLAM is an advanced real-time SLAM system that combines Fourier-based Gaussian Splatting with sparse and dense map fusion for high-performance 3D reconstruction and localization.

• NVIDIA GPU with CUDA 11.8 support
• Conda package manager

1. Clone the repository:

   git clone https://github.com/3DV-Coder/FGS-SLAM.git --recursive
   cd FGS-SLAM
   

2. Create conda environment:

   conda create -n fgsslam python==3.9
   conda activate fgsslam

3. Install PyTorch:

   conda install pytorch==2.0.0 torchvision==0.15.0 torchaudio==2.0.0 pytorch-cuda=11.8 -c pytorch -c nvidia

4. Install requirements:

   pip install -r requirements.txt
   conda install -c conda-forge python-pcl

5. Install submodules:

   pip install submodules/diff-gaussian-rasterization
   pip install submodules/simple-knn
   
   cd submodules/fast_gicp
   mkdir -p build && cd build
   cmake .. && make
   cd ..
   python setup.py install --user

bash download_replica.sh

# Modify the directory structure
python reorganize_replica.py ./dataset/Replica	

bash download_tum.sh

bash replica_unlimit.sh

bash tum_unlimit.sh

1. Install dependencies:

   sudo apt install -y libglew-dev libassimp-dev libboost-all-dev libgtk-3-dev libopencv-dev libglfw3-dev libavdevice-dev libavcodec-dev libeigen3-dev libxxf86vm-dev libembree-dev

2. Build the viewer:

   cd SIBR_viewers
   cmake -Bbuild . -DCMAKE_BUILD_TYPE=Release # add -G Ninja to build faster
   cmake --build build -j24 --target install

3. Run the system:

   python -W ignore fgs_slam.py --dataset_path dataset/Replica/office0 --verbose

4. In another terminal:

   cd SIBR_viewers
   ./install/bin/SIBR_remoteGaussian_app --rendering-size 1280 720

python -W ignore fgs_slam.py --rerun_viewer

We extend our gratitude to the following projects and their contributors for their foundational work and inspiration: