[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-79869":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":10,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":24,"hasPages":22,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":26,"readmeContent":27,"aiSummary":28,"trendingCount":16,"starSnapshotCount":16,"syncStatus":29,"lastSyncTime":30,"discoverSource":31},79869,"HumanEgo","TX-Leo\u002FHumanEgo","TX-Leo","HumanEgo: Zero-Shot Robot Learning from Minutes of Human Egocentric Videos","https:\u002F\u002Fhumanego-ai.github.io",null,"Python",285,32,3,4,0,12,54,139,48,4.56,false,"main",true,[],"2026-06-12 02:03:55","\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fhumanego-ai.github.io\">\n    \u003Cimg src=\"assets\u002Ftitle\u002Fhero.png\" alt=\"HumanEgo — Zero-Shot Robot Learning from Minutes of Human Egocentric Videos\" width=\"100%\" \u002F>\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Ftx-leo.github.io\">Zhi (Leo) Wang\u003C\u002Fa>  · \n  \u003Ca href=\"https:\u002F\u002Fbottle101.github.io\u002F\">Botao He\u003C\u002Fa>  · \n  \u003Ca href=\"https:\u002F\u002Fcolinyu1.github.io\u002F\">Kelin Yu\u003C\u002Fa>  · \n  \u003Ca href=\"https:\u002F\u002Fsjlee.cc\u002F\">Seungjae Lee\u003C\u002Fa>  · \n  \u003Ca href=\"https:\u002F\u002Fruohangao.github.io\u002F\">Ruohan Gao\u003C\u002Fa>  · \n  \u003Ca href=\"https:\u002F\u002Ffurong-huang.com\u002F\">Furong Huang\u003C\u002Fa>  · \n  \u003Ca href=\"https:\u002F\u002Frobotics.umd.edu\u002Fclark\u002Ffaculty\u002F350\u002FYiannis-Aloimonos\">Yiannis Aloimonos\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fhumanego-ai.github.io\">\u003Cimg src=\"assets\u002Ftitle\u002Fbtn_website.png\" alt=\"Website\" height=\"60\" \u002F>\u003C\u002Fa>\n   \n  \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2605.24934\">\u003Cimg src=\"assets\u002Ftitle\u002Fbtn_paper.png\" alt=\"Paper\" height=\"60\" \u002F>\u003C\u002Fa>\n   \n  \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.24934\">\u003Cimg src=\"assets\u002Ftitle\u002Fbtn_arxiv.png\" alt=\"arXiv\" height=\"60\" \u002F>\u003C\u002Fa>\n   \n  \u003Ca href=\"https:\u002F\u002Fyoutu.be\u002FpdL46diijuY\">\u003Cimg src=\"assets\u002Ftitle\u002Fbtn_video.png\" alt=\"Video\" height=\"60\" \u002F>\u003C\u002Fa>\n   \n  \u003Ca href=\"#bibtex\">\u003Cimg src=\"assets\u002Ftitle\u002Fbtn_bibtex.png\" alt=\"BibTeX\" height=\"60\" \u002F>\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fteaser.gif\" alt=\"HumanEgo teaser\" width=\"100%\" \u002F>\n\u003C\u002Fp>\n\n---\n\n## Installation\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FTX-Leo\u002FHumanEgo.git\ncd HumanEgo\nconda create -n humanego python=3.11 -y\nconda activate humanego\nbash setup.sh\n```\n\nThis installs PyTorch (with CUDA), the vision foundation models we use\n(SAM 2, Grounding DINO, CoTracker, Orient-Anything V2), and the hand-tracking\nmethods (MediaPipe, WiLoR, HaMeR).\n\n---\n\n## Data Collection\n\nSee [`datacollection\u002FREADME_data_collection.md`](datacollection\u002FREADME_data_collection.md)\nfor the end-to-end guide on recording your own Project Aria data and running\nMPS (SLAM + hand tracking) on it.\n\n> **Dataset release — coming soon.** We will be sharing the full HumanEgo\n> dataset (raw Aria recordings + MPS-processed annotations) publicly. Stay\n> tuned.\n\n---\n\n## Preprocess\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fdata_collection.webp\" alt=\"HumanEgo preprocessing visualization\" width=\"100%\" \u002F>\n\u003C\u002Fp>\n\n> **TODO** — documentation coming soon.\n\n---\n\n## Training\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Farchitecture.webp\" alt=\"HumanEgo training architecture\" width=\"100%\" \u002F>\n\u003C\u002Fp>\n\n```bash\npython -m training.FlowMatchingTrainer --task \"YOUR_TASK\" --use_cfg --job \"YOUR_JOB\"\n```\n\n`--task` selects a folder under `cfg\u002Ftraining\u002F` (e.g. `serve_bread`) and\n`--job` selects a YAML inside it (e.g. `HumanEgo`, resolving to\n`cfg\u002Ftraining\u002Fserve_bread\u002FHumanEgo.yaml`).\n\n---\n\n## Inference\n\n> **TODO** — documentation coming soon.\n\n---\n\n## TODO\n\nWe are actively releasing the following — check back soon.\n\n- [ ] Release a 3-minute quick-start tutorial\n- [ ] Release a sample human-egocentric dataset (for end-to-end testing)\n- [ ] Release a pretrained model (for inference demo)\n- [ ] Release documentation for **Preprocessing**\n- [ ] Release documentation for **Training**\n- [ ] Release documentation for **Inference**\n\n---\n\n## Acknowledgements\n\nThis project builds on excellent open-source work, including\n[Project Aria](https:\u002F\u002Fwww.projectaria.com\u002F) (Gen 1 glasses &\n[MPS](https:\u002F\u002Ffacebookresearch.github.io\u002Fprojectaria_tools\u002Fdocs\u002Fintro)),\n[Trossen Arm](https:\u002F\u002Fwww.trossenrobotics.com\u002F),\n[CoTracker3](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Fco-tracker),\n[Grounding DINO](https:\u002F\u002Fgithub.com\u002FIDEA-Research\u002FGroundingDINO),\n[SAM 2](https:\u002F\u002Fgithub.com\u002Ffacebookresearch\u002Fsam2),\n[HaMeR](https:\u002F\u002Fgithub.com\u002Fgeopavlakos\u002Fhamer),\n[WiLoR](https:\u002F\u002Fgithub.com\u002Frolpotamias\u002FWiLoR),\n[MediaPipe](https:\u002F\u002Fgithub.com\u002Fgoogle-ai-edge\u002Fmediapipe),\n[LaMa](https:\u002F\u002Fgithub.com\u002Fadvimman\u002Flama),\nand [Orient-Anything](https:\u002F\u002Fgithub.com\u002FSpatialVision\u002FOrient-Anything).\n\n---\n\n\u003Ch2 id=\"bibtex\">BibTeX\u003C\u002Fh2>\n\n```bibtex\n@misc{humanego2026,\n  title         = {HumanEgo: Zero-Shot Robot Learning from Minutes of Human Egocentric Videos},\n  author        = {Wang, Zhi and He, Botao and Yu, Kelin and Lee, Seungjae and Gao, Ruohan and Huang, Furong and Aloimonos, Yiannis},\n  year          = {2026},\n  eprint        = {2605.24934},\n  archivePrefix = {arXiv},\n  primaryClass  = {cs.RO}\n}\n```\n","HumanEgo 是一个基于人类第一视角视频实现零样本机器人学习的项目。它通过几分钟的人类第一视角视频，使机器人能够学习并模仿人类的行为。该项目使用了多种先进的视觉基础模型（如SAM 2、Grounding DINO等）和手部跟踪方法（如MediaPipe），以处理复杂的视觉信息，并通过FlowMatchingTrainer进行训练。适合于需要快速部署机器人模仿人类行为的应用场景，比如家庭服务机器人或工业自动化领域。",2,"2026-06-11 03:58:22","CREATED_QUERY"]