[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72590":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":25,"hasPages":23,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":32,"readmeContent":33,"aiSummary":34,"trendingCount":16,"starSnapshotCount":16,"syncStatus":35,"lastSyncTime":36,"discoverSource":37},72590,"MonoGS","muskie82\u002FMonoGS","muskie82","[CVPR'24 Highlight & Best Demo Award] Gaussian Splatting SLAM","https:\u002F\u002Frmurai.co.uk\u002Fprojects\u002FGaussianSplattingSLAM\u002F",null,"Python",2104,223,23,100,0,3,8,29,9,29.05,"Other",false,"main",true,[27,28,29,30,31],"computer-vision","cvpr2024","gaussian-splatting","robotics","slam","2026-06-12 02:03:05","[comment]: \u003C> (# Gaussian Splatting SLAM)\n\n\u003C!-- PROJECT LOGO -->\n\n\u003Cp align=\"center\">\n\n  \u003Ch1 align=\"center\"> Gaussian Splatting SLAM\n  \u003C\u002Fh1>\n  \u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Fmuskie82.github.io\u002F\">\u003Cstrong>*Hidenobu Matsuki\u003C\u002Fstrong>\u003C\u002Fa>\n    ·\n    \u003Ca href=\"https:\u002F\u002Frmurai.co.uk\u002F\">\u003Cstrong>*Riku Murai\u003C\u002Fstrong>\u003C\u002Fa>\n    ·\n    \u003Ca href=\"https:\u002F\u002Fwww.imperial.ac.uk\u002Fpeople\u002Fp.kelly\u002F\">\u003Cstrong>Paul H.J. Kelly\u003C\u002Fstrong>\u003C\u002Fa>\n    ·\n    \u003Ca href=\"https:\u002F\u002Fwww.doc.ic.ac.uk\u002F~ajd\u002F\">\u003Cstrong>Andrew J. Davison\u003C\u002Fstrong>\u003C\u002Fa>\n  \u003C\u002Fp>\n  \u003Cp align=\"center\">(* Equal Contribution)\u003C\u002Fp>\n\n  \u003Ch3 align=\"center\"> CVPR 2024 (Highlight)\u003C\u002Fh3>\n\n\n\n[comment]: \u003C> (  \u003Ch2 align=\"center\">PAPER\u003C\u002Fh2>)\n  \u003Ch3 align=\"center\">\u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.06741\">Paper\u003C\u002Fa> | \u003Ca href=\"https:\u002F\u002Fyoutu.be\u002Fx604ghp9R_Q?si=nYoWr8h2Xh-6L_KN\">Video\u003C\u002Fa> | \u003Ca href=\"https:\u002F\u002Frmurai.co.uk\u002Fprojects\u002FGaussianSplattingSLAM\u002F\">Project Page\u003C\u002Fa>\u003C\u002Fh3>\n  \u003Cdiv align=\"center\">\u003C\u002Fdiv>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"\">\n    \u003Cimg src=\".\u002Fmedia\u002Fteaser.gif\" alt=\"teaser\" width=\"100%\">\n  \u003C\u002Fa>\n  \u003Ca href=\"\">\n    \u003Cimg src=\".\u002Fmedia\u002Fgui.jpg\" alt=\"gui\" width=\"100%\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\u003Cp align=\"center\">\nThis software implements dense SLAM system presented in our paper \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.06741\">Gaussian Splatting SLAM\u003C\u002Fa> in CVPR'24.\nThe method demonstrates the first monocular SLAM solely based on 3D Gaussian Splatting (left), which also supports Stereo\u002FRGB-D inputs (middle\u002Fright).\n\u003C\u002Fp>\n\u003Cbr>\n\n# Note\n- In an academic paper, please refer to our work as **Gaussian Splatting SLAM** or **MonoGS** for short (this repo's name) to avoid confusion with other works.\n- Differential Gaussian Rasteriser with camera pose gradient computation is available [here](https:\u002F\u002Fgithub.com\u002Frmurai0610\u002Fdiff-gaussian-rasterization-w-pose.git).\n- **[New]** Speed-up version of our code is available in `dev.speedup` branch, It achieves up to 10fps on monocular fr3\u002Foffice sequence while keeping consistent performance (tested on RTX4090\u002Fi9-12900K). The code will be merged into the main branch after further refactoring and testing.\n\n# Getting Started\n## Installation\n```\ngit clone https:\u002F\u002Fgithub.com\u002Fmuskie82\u002FMonoGS.git --recursive\ncd MonoGS\n```\nSetup the environment.\n\n```\nconda env create -f environment.yml\nconda activate MonoGS\n```\nDepending on your setup, please change the dependency version of pytorch\u002Fcudatoolkit in `environment.yml` by following [this document](https:\u002F\u002Fpytorch.org\u002Fget-started\u002Fprevious-versions\u002F).\n\nOur test setup were:\n- Ubuntu 20.04: `pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.6`\n- Ubuntu 18.04: `pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3`\n\n## Quick Demo\n```\nbash scripts\u002Fdownload_tum.sh\npython slam.py --config configs\u002Fmono\u002Ftum\u002Ffr3_office.yaml\n```\nYou will see a GUI window pops up.\n\n## Downloading Datasets\nRunning the following scripts will automatically download datasets to the `.\u002Fdatasets` folder.\n### TUM-RGBD dataset\n```bash\nbash scripts\u002Fdownload_tum.sh\n```\n\n### Replica dataset\n```bash\nbash scripts\u002Fdownload_replica.sh\n```\n\n### EuRoC MAV dataset\n```bash\nbash scripts\u002Fdownload_euroc.sh\n```\n\n\n\n## Run\n### Monocular\n```bash\npython slam.py --config configs\u002Fmono\u002Ftum\u002Ffr3_office.yaml\n```\n\n### RGB-D\n```bash\npython slam.py --config configs\u002Frgbd\u002Ftum\u002Ffr3_office.yaml\n```\n\n```bash\npython slam.py --config configs\u002Frgbd\u002Freplica\u002Foffice0.yaml\n```\nOr the single process version as\n```bash\npython slam.py --config configs\u002Frgbd\u002Freplica\u002Foffice0_sp.yaml\n```\n\n\n### Stereo (experimental)\n```bash\npython slam.py --config configs\u002Fstereo\u002Feuroc\u002Fmh02.yaml\n```\n\n## Live demo with Realsense\nFirst, you'll need to install `pyrealsense2`.\nInside the conda environment, run:\n```bash\npip install pyrealsense2\n```\nConnect the realsense camera to the PC on a **USB-3** port and then run:\n```bash\npython slam.py --config configs\u002Flive\u002Frealsense.yaml\n```\nWe tested the method with [Intel Realsense d455](https:\u002F\u002Fwww.mouser.co.uk\u002Fnew\u002Fintel\u002Fintel-realsense-depth-camera-d455\u002F). We recommend using a similar global shutter camera for robust camera tracking. Please avoid aggressive camera motion, especially before the initial BA is performed. Check out [the first 15 seconds of our YouTube video](https:\u002F\u002Fyoutu.be\u002Fx604ghp9R_Q?si=S21HgeVTVfNe0BVL) to see how you should move the camera for initialisation. We recommend to use the code in `dev.speed-up` branch for live demo.\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"\">\n    \u003Cimg src=\".\u002Fmedia\u002Frealsense.png\" alt=\"teaser\" width=\"50%\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n# Evaluation\n\u003C!-- To evaluate the method, please run the SLAM system with `save_results=True` in the base config file. This setting automatically outputs evaluation metrics in wandb and exports log files locally in save_dir. For benchmarking purposes, it is recommended to disable the GUI by setting `use_gui=False` in order to maximise GPU utilisation. For evaluating rendering quality, please set the `eval_rendering=True` flag in the configuration file. -->\nTo evaluate our method, please add `--eval` to the command line argument:\n```bash\npython slam.py --config configs\u002Fmono\u002Ftum\u002Ffr3_office.yaml --eval\n```\nThis flag will automatically run our system in a headless mode, and log the results including the rendering metrics.\n\n# Reproducibility\nThere might be minor differences between the released version and the results in the paper. Please bear in mind that multi-process performance has some randomness due to GPU utilisation.\nWe run all our experiments on an RTX 4090, and the performance may differ when running with a different GPU.\n\n# Acknowledgement\nThis work incorporates many open-source codes. We extend our gratitude to the authors of the software.\n- [3D Gaussian Splatting](https:\u002F\u002Fgithub.com\u002Fgraphdeco-inria\u002Fgaussian-splatting)\n- [Differential Gaussian Rasterization\n](https:\u002F\u002Fgithub.com\u002Fgraphdeco-inria\u002Fdiff-gaussian-rasterization)\n- [SIBR_viewers](https:\u002F\u002Fgitlab.inria.fr\u002Fsibr\u002Fsibr_core)\n- [Tiny Gaussian Splatting Viewer](https:\u002F\u002Fgithub.com\u002Flimacv\u002FGaussianSplattingViewer)\n- [Open3D](https:\u002F\u002Fgithub.com\u002Fisl-org\u002FOpen3D)\n- [Point-SLAM](https:\u002F\u002Fgithub.com\u002Feriksandstroem\u002FPoint-SLAM)\n\n# License\nMonoGS is released under a **LICENSE.md**. For a list of code dependencies which are not property of the authors of MonoGS, please check **Dependencies.md**.\n\n# Citation\nIf you found this code\u002Fwork to be useful in your own research, please considering citing the following:\n\n```bibtex\n@inproceedings{Matsuki:Murai:etal:CVPR2024,\n  title={{G}aussian {S}platting {SLAM}},\n  author={Hidenobu Matsuki and Riku Murai and Paul H. J. Kelly and Andrew J. Davison},\n  booktitle={Proceedings of the IEEE\u002FCVF Conference on Computer Vision and Pattern Recognition},\n  year={2024}\n}\n\n```\n\n\n\n\n\n\n\n\n\n\n\n\n\n","MonoGS 是一个基于高斯点云的单目 SLAM 系统，荣获 CVPR 2024 最佳演示奖。其核心功能在于通过3D高斯点云实现稠密SLAM，并支持单目、立体和RGB-D输入。技术上，该项目使用Python开发，具有差异化的高斯光栅化器以及相机姿态梯度计算能力。此外，项目还提供了一个加速版本，在RTX4090\u002Fi9-12900K硬件配置下能达到最高10fps的处理速度。MonoGS特别适用于需要高精度定位与建图的机器人导航场景及计算机视觉研究领域。",2,"2026-06-11 03:42:42","high_star"]