VGGT-SLAM: Dense RGB SLAM Optimized on the SL(4) Manifold
Make sure the following dependencies are installed before building the project:
sudo apt-get install git python3-pip libboost-all-dev cmake gcc g++ unzip
Then, clone VGGT-SLAM:
git clone https://github.com/MIT-SPARK/VGGT-SLAM
cd VGGT-SLAM
conda create -n vggt-slam python=3.11
conda activate vggt-slam
This step will automatically download all 3rd party packages including VGGT. More details on the license for VGGT can be found here.
./setup.sh
run python main.py --image_folder /path/to/image/folder --max_loops 1 --vis_map replacing the image path with your folder of images.
This will create a visualization in viser which shows the incremental construction of the map.
As an example, we provide a folder of test images in office_loop.zip which will generate the following map. Using the default parameters will
result in a single loop closure towards the end of the trajectory. Unzip the folder and set its path as the arguments for --image_folder, e.g.,
unzip office_loop.zip
and then run the below command:
python3 main.py --image_folder office_loop --max_loops 1 --vis_map
Running in the default SL(4) mode on this folder will show significant drift in the projective degrees of freedom before the loop closure, and the drift will be corrected after the loop closure. You may notice drift in other scenes as well if the system goes too long without a loop closure. We are actively working on an upgraded VGGT-SLAM that will have significantly reduced drift and other major updates so stay tuned!
To quickly collect a test on a custom dataset, you can record a trajectory with a cell phone and convert the MOV file to a folder of images with:
mkdir <desired_location>/img_folder
And then, run the command below:
ffmpeg -i /path/to/video.MOV -vf "fps=10" <desired_location>/img_folder/frame_%04d.jpg
See main.py or run --help from main.py to view all parameters.
We use SL(4) mode by default, and Sim(3) mode can be enabled with --use_sim3. Sim(3) mode will generally have less drift than SL(4) but will not always
be sufficient for alignment (see paper for in depth discussion on the advantages of SL(4)).
To automatically run evaluation on TUM and 7-Scenes datasets, first install the datasets using the provided download instructions from MASt3R-SLAM. Set the download location of MASt3R-SLAM by setting abs_dir in the bash scripts /evals/eval_tum.sh and /evals/eval_7scenes.sh
To run on TUM, run ./evals/eval_tum.sh <w> and then run python evals/process_logs_tum.py --submap_size <w> to analyze and print the results, where w is
the submap size, for example:
./evals/eval_tum.sh 32
python evals/process_logs_tum.py --submap_size 32
To run on 7-Scenes, run ./evals/eval_7scenes.sh <w> and then run python evals/process_logs_7scenes.py <w> to analyze and print the results, where w is
the submap size, for example:
./evals/eval_7scenes.sh 32
python evals/process_logs_7scenes.py 32
By default, ever scene will be run for 5 trials, this can be changed inside the bash scripts.
To visualize the maps as they being constructed, inside the bash scripts add --vis_map. This will update the viser map each time the submap is updated.
- August 2025: SL(4) optimization is integrated into the official GTSAM repo
This work is supported in part by the NSF Graduate Research Fellowship Program under Grant 2141064, the ONR RAPID program, and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. RS-2024-00461409). The authors would like to gratefully acknowledge Riku Murai for assisting us with benchmarking.
If our code is helpful, please cite our paper as follows:
@article{maggio2025vggt,
title={VGGT-SLAM: Dense RGB SLAM Optimized on the SL (4) Manifold},
author={Maggio, Dominic and Lim, Hyungtae and Carlone, Luca},
journal={arXiv preprint arXiv:2505.12549},
year={2025}
}
