{"id":21822043,"url":"https://github.com/longxiang-ai/human101","last_synced_at":"2026-01-03T23:45:31.557Z","repository":{"id":211688767,"uuid":"729736499","full_name":"longxiang-ai/Human101","owner":"longxiang-ai","description":"The official implementation of \"Human101: Training 100+FPS Human Gaussians in 100s from 1 View\".","archived":false,"fork":false,"pushed_at":"2023-12-27T07:56:54.000Z","size":85369,"stargazers_count":111,"open_issues_count":1,"forks_count":1,"subscribers_count":25,"default_branch":"main","last_synced_at":"2025-01-26T07:41:20.314Z","etag":null,"topics":["3d","3d-reconstruction","3d-vision","computer-vision","digital-human","gaussian","gaussian-splatting","human-reconstruction","monocular","real-time","real-time-rendering","view-synthesis"],"latest_commit_sha":null,"homepage":"https://longxiang-ai.github.io/Human101/","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/longxiang-ai.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2023-12-10T07:22:26.000Z","updated_at":"2024-12-10T05:42:33.000Z","dependencies_parsed_at":"2024-01-24T12:05:46.689Z","dependency_job_id":"9837b8da-770d-471c-a52a-942c2053ddaa","html_url":"https://github.com/longxiang-ai/Human101","commit_stats":{"total_commits":6,"total_committers":3,"mean_commits":2.0,"dds":"0.33333333333333337","last_synced_commit":"7ef03485ad4eb4c848b24a5a8ab1a699d73ff002"},"previous_names":["longxiang-ai/human101"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/longxiang-ai%2FHuman101","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/longxiang-ai%2FHuman101/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/longxiang-ai%2FHuman101/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/longxiang-ai%2FHuman101/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/longxiang-ai","download_url":"https://codeload.github.com/longxiang-ai/Human101/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":244787014,"owners_count":20510041,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["3d","3d-reconstruction","3d-vision","computer-vision","digital-human","gaussian","gaussian-splatting","human-reconstruction","monocular","real-time","real-time-rendering","view-synthesis"],"created_at":"2024-11-27T17:12:44.809Z","updated_at":"2026-01-03T23:45:31.525Z","avatar_url":"https://github.com/longxiang-ai.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n# \u003cb\u003eHuman101\u003c/b\u003e: Training 100+FPS Human Gaussians in 100s from 1 View\n\n[Mingwei Li](https://github.com/longxiang-ai), [Jiachen Tao](https://github.com/tlxhlll), [Zongxin Yang](https://z-x-yang.github.io), [Yi Yang](https://scholar.google.com/citations?user=RMSuNFwAAAAJ)\u003csup\u003e\\*\u003c/sup\u003e\n\n\u003csup\u003e\\*\u003c/sup\u003eCorresponding author\n\nReLER, CCAI, Zhejiang University\n\n[[Project Page](https://longxiang-ai.github.io/Human101/)] [[ArXiv](https://arxiv.org/abs/2312.15258)] [[Supplementary Material](https://arxiv.org/abs/2312.15258)]\n\n\u003c/div\u003e\n\n## Introduction\n\nWe propose **Human101**, a fast 1-view human reconstruction framework. Human101 is able to train 3D Gaussians in 100 seconds and render 1024-resolution images at 60+ FPS, **without necessitating the pre-storage of per-frame Gaussian attributes**. The pipeline of Human101 is shown as follows:\n![pipeline](./assets/pipeline.png)\n\n## Abstract\n\nReconstructing the human body from single-view videos plays a pivotal role in the virtual reality domain. One prevalent application scenario necessitates the rapid reconstruction of high-fidelity 3D digital humans while simultaneously ensuring real-time rendering and interaction. Existing methods often struggle to fulfill both requirements. In this paper, we introduce Human101, a novel framework adept at producing high-fidelity dynamic 3D human reconstructions from 1-view videos by training 3D Gaussians in 100 seconds and rendering in 100+ FPS. Our method leverages the strengths of 3D Gaussian Splatting, which provides an explicit and efficient representation of 3D humans. Standing apart from prior NeRF-based systems, Human101 ingeniously applies a Human-centric Forward Gaussian Animation to deform the parameters of 3D Gaussians, thereby enhancing rendering speed (i.e., rendering 1024-resolution images at an impressive 60+ FPS and rendering 512-resolution images at 100+ FPS). Experimental results indicate that our approach substantially eclipses current methods, clocking up to a 10 $ \\times $ surge in frames per second and delivering comparable or superior rendering quality.\n\n## News\n\n- [2023/12/27] We release paper on [arXiv](https://arxiv.org/abs/2312.15258) and the [project page](https://longxiang-ai.github.io/Human101/).\n\n## TODO list\n\n- [√] Release demos \u0026 project page\n- [ ] Release code\n\n## Acknowledgement\n\nOur implementation is mainly based on [3D Gaussian Splatting](https://github.com/graphdeco-inria/gaussian-splatting) , [Instant-nvr](https://github.com/zju3dv/instant-nvr), [InstantAvatar](https://github.com/tijiang13/InstantAvatar)\nand the following open-source projects:\n\n- [ECON](https://github.com/YuliangXiu/ECON)\n- [GTA](https://github.com/River-Zhang/GTA)\n- [EasyMoCap](https://github.com/zju3dv/EasyMocap/)\n- [DreamGaussian](https://github.com/dreamgaussian/dreamgaussian)\n\nAnd many thanks to the authors of [GTA](https://github.com/River-Zhang/GTA) and [TransHuman](https://github.com/pansanity666/TransHuman) for discussing some details about the implementation.\n\nMore related papers about 3D avatars: [Awesome-3D-Avatars](https://github.com/pansanity666/Awesome-Avatars).\n\n## Citation\n\nIf you find this code useful for your research, please consider citing:\n\n```bibtex\n@misc{li2023human101,\n      title={Human101: Training 100+FPS Human Gaussians in 100s from 1 View},\n      author={Mingwei Li and Jiachen Tao and Zongxin Yang and Yi Yang},\n      year={2023},\n      eprint={2312.15258},\n      archivePrefix={arXiv},\n      primaryClass={cs.CV}\n}\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flongxiang-ai%2Fhuman101","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Flongxiang-ai%2Fhuman101","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flongxiang-ai%2Fhuman101/lists"}