Blomega Data Refinery / registry
AI training datasets and whether you may use them commercially
73 records, built 2026-09-22. Machine-readable: JSON and JSON-LD.
| Name | Organisation | Hours | Episodes | Licence | Commercial use | Access | Grade | Source |
|---|---|---|---|---|---|---|---|---|
| 10Kh-RealOmin-OpenData | GenRobot | not established | not established | CC-BY-SA-4.0 | not established | auto | B | source |
| AetheRock demonstration set | Shanghai Jiao Tong University, Ant Group | 2.32 | 1275 | No licence at any layer: the HuggingFace card YAML has only pretty_name and tags, the README states no terms, and the re... | not established | not established | A | source |
| AgiBot World | AgiBot | 2976 | 1001552 | CC BY-NC (check) | no | not established | B | source |
| AXIS Franka Dataset (Axis Sim Dataset V1) | Axis Robotics | not established | 50000 | other (gated, non-commercial academic/education only) | no | not established | B | source |
| Bench2Dex | Shanghai Jiao Tong University | not established | 1300 | MIT | not established | not established | B | source |
| Beyond Gestures HOM (hand-object manipulation) | Meta Reality Labs Research | 19.3 | 55 | not released | no | not established | B | source |
| Beyond Gestures HP + FF (single participant) | Meta Reality Labs Research | 12.7 | 38 | not released | no | not established | B | source |
| BioDCASE 2026 Task 4: ATBFL | BioDCASE, Australian Antarctic Data Centre | 21.7 | not established | CC-BY-4.0 | yes | not established | B | source |
| BioDCASE 2026 Task 4: BirdSet subsets (HSN, POW, UHH) | BioDCASE, BirdSet | not established | not established | CC-BY-4.0 | not established | not established | B | source |
| Blomega GX-1 capture | Blomega | not established | not established | commercial | yes | not established | B | source |
| BridgeData V2 | UC Berkeley | 333.9 | 60096 | CC BY 4.0 | yes | not established | A | source |
| c4 | Allen Institute for AI | not established | 365000000 | ODC-By-1.0 | yes | no | B | source |
| California Data Broker Registry 2026 (first-party count, Talika Broker-to-Model Count) | California Privacy Protection Agency | not established | not established | 'In general, information presented on this website, unless otherwise indicated, is considered in the public domain. It m... | yes | not established | B | source |
| Community Notes public data (2026-09-20 snapshot) | X | not established | not established | Community Notes data licence | not established | no | B | source |
| Compliance for Free bilateral wiping demonstrations (Franka Research 3) | IRI | not established | 96 | code and data 'coming soon' as of 2026-09-19 | not established | not established | B | source |
| dclm-baseline-1.0 | ML Foundations | not established | not established | CC-BY-4.0 | not established | no | B | source |
| DexCap | Stanford | not established | 787 | CC-BY-4.0 | yes | not established | B | source |
| DICOM Medical Imaging Dataset with Clinical Reports | BLOMEGA | not established | not established | OTS (healthcare DICOM paired with clinical reports) | yes | not established | B | source |
| Doctor Appointment Emails | BLOMEGA | not established | 7470 | OTS (booking emails: confirmation/cancellation/modification) | yes | not established | B | source |
| Domestic Animal Behavior | BLOMEGA | not established | 408000 | OTS (408K clips) | yes | not established | B | source |
| Dreaming Contact Dataset (v1) | GRASP Lab, University of Pennsylvania | not established | 200 | not established | not established | no | B | source |
| DROID | Stanford | 350 | 76000 | CC-BY-4.0 | yes | not established | A | source |
| DYNA-2 corpus | Dyna Robotics | 1000000 | not established | proprietary (no public release) | no | not established | B | source |
| Ego-Exo4D | Meta | 1286 | 5035 | Bespoke Ego-Exo4D licence agreement, 12 licensors: commercial product development allowed for enumerated purposes; no su... | yes | not established | A | source |
| Ego2Robot synthesized robot data | Qwen Team, Renmin University AIM3, ShanghaiTech, BIGAI | 18561 | not established | not released (project page code link is '#') | no | not established | B | source |
| Ego4D | Meta | 3670 | not established | Ego4D license | yes | not established | B | source |
| EgoDex | Apple | 829 | not established | CC BY-NC-ND (non-commercial, no derivatives), stated in the apple/ml-egodex README; LeRobot-format mirrors on Hugging Face carry cc-by-nc... | no | not established | B | source |
| EgoScale | UT Austin RPL | 20854 | not established | No dataset licence published: the paper states no terms and offers no download, no repository or card exists for it, and... | not established | not established | B | source |
| EgoVerse | Georgia Tech, Stanford, UC San Diego, ETH, Massachusetts Institute of Technology, Meta, Scale | 1362 | 80000 | MIT | yes | not established | B | source |
| English Animation Videos | BLOMEGA | 8500 | not established | OTS, en | yes | not established | B | source |
| English-French Translation Pairs | BLOMEGA | not established | 22500000 | ODbL-1.0 | yes | not established | B | source |
| EPIC-KITCHENS | University of Bristol | 100 | 700 | CC BY-NC | no | not established | A | source |
| Europarl Parallel Corpus | BLOMEGA | not established | 2000000 | free; 2M pairs across 21 EU languages | yes | not established | B | source |
| FarsTail | Iranian NLP community | not established | not established | Apache-2.0 | yes | not established | B | source |
| FineFineWeb | M-A-P | not established | not established | Apache-2.0 | not established | no | B | source |
| Food Videos | BLOMEGA | 5000 | not established | OTS, en | yes | not established | B | source |
| ForceBand EMG2Force pretraining dataset | Amazon FAR, University of Maryland, Johns Hopkins | 10 | not established | unreleased, listed as under preparation | no | not established | B | source |
| GAPA (Gender Associations of Physical Attributes) | University of California, Los Angeles | not established | not established | MIT | yes | not established | B | source |
| GIFT glove demonstration corpus (cup grasp-and-hold) | Tzah Sarusi | not established | 73 | not established | not established | no | B | source |
| HumanTouch (DexTouch-WM human interaction corpus) | HKUST, Xspark AI, Tsinghua, PKU, HKU | 100 | not established | not released as of 2026-09-19 | no | not established | B | source |
| Industrial Electric Motor Thermography | BLOMEGA | not established | 5800 | OTS (5.8K thermographic images for predictive maintenance) | yes | not established | B | source |
| LESS released selections (less_data) | Princeton NLP | not established | not established | not established | not established | no | B | source |
| LLaVA-OneVision-1.5-Mid-Training-85M | MVP Lab | not established | not established | Apache-2.0 | not established | no | B | source |
| Meeting Recordings | BLOMEGA | 99500 | not established | OTS, consent-verified | yes | not established | B | source |
| Mobile ALOHA | Stanford | not established | 276 | MIT | yes | not established | B | source |
| Multi-Person Social Interaction | BLOMEGA | not established | 478000 | OTS (478K clips) | yes | not established | B | source |
| Multilingual Conversations | BLOMEGA | 976000 | not established | consent-verified, license-clear (OTS) | yes | not established | B | source |
| Multilingual UN Parallel Corpus | BLOMEGA | not established | 11000000 | free; 11M pairs across 6 official languages | yes | not established | B | source |
| Multimodal Dataset for Household Robotics | BLOMEGA | not established | not established | OTS (3D and Lidar multimodal household robotics data) | yes | not established | B | source |
| NSURL-2019 Persian NER | NSURL | not established | not established | No licence published: the task organisers' repository carries no LICENSE file and the README states none; training part ... | not established | not established | B | source |
| objaverse | Allen Institute for AI | not established | not established | README License section: the dataset as a whole is ODC-By v1.0, but individual objects carry their own Creative Commons l... | no | no | B | source |
| Objective Projection (v7.2) | Levent Bulut | not established | not established | cc-by-nc-nd-4.0 | no | not established | C | source |
| Open X-Embodiment | Google DeepMind | 2000 | 1000000 | CC-BY-4.0 | yes | not established | A | source |
| Open-AoE (OpenAoE-2000h) | Ant Group, inclusionAI | 2000 | not established | Open-AoE Dataset License v1.0 (commercial with attribution) for video+annotations; hands.npz and camera_traj.npz additionally subject to ... | yes | not established | B | source |
| OpenEAI-Dataset | OpenEAI | not established | not established | not established | not established | no | C | source |
| OSMO human demonstration set | Meta, University of Pennsylvania, University of Michigan | 2 | 140 | No licence and no release: GitHub's licence detection on jessicayin/osmo_tactile_glove returns null, the repository root... | not established | not established | A | source |
| PhysicalAI-Robotics-GR00T-X-Embodiment-Sim | NVIDIA | not established | not established | CC-BY-4.0 | not established | no | B | source |
| Pose-Related Artifact glove dataset (arXiv 2607.22964) | Cornell, Meta, University of Washington | 13.8 | not established | not released | no | not established | B | source |
| Project Kitchen gameplay corpus | Hong Kong University of Science and Technology | 10 | not established | unreleased, announced for release upon acceptance | no | yes | B | source |
| Public Discourse Corpus (PDC) | Institute of Computing Technology, Chinese Academy of Sciences | not established | not established | not established | not established | not established | B | source |
| Real Job Interview Audio (24+ Accents) | BLOMEGA | 10000 | not established | consented (for ASR/LLMs), 100 GB, 24+ accents | not established | not established | B | source |
| RH20T | Shanghai Jiao Tong University | not established | 110000 | research | no | not established | B | source |
| RoboCasa | University of Texas at Austin, NVIDIA | 420 | not established | CC-BY-4.0 | yes | not established | B | source |
| RoboNet | not established | 900 | 162000 | CC-BY-4.0 | yes | not established | B | source |
| RT-1 dataset | not established | 130000 | research | no | not established | B | source | |
| Seoul pedestrian-level sidewalk accessibility benchmark | MIT Senseable City Lab | not established | not established | not established | not established | not established | C | source |
| Something-Something V2 | TwentyBN, Qualcomm | not established | 220847 | research | no | not established | B | source |
| STAR 200-hour bimanual dexterous dataset | Jianlan Luo | 200 | 10576 | not established | not established | not established | B | source |
| Thai Media Content Videos | BLOMEGA | 3100 | not established | OTS, th | yes | not established | B | source |
| UMI | Stanford, Columbia | not established | not established | MIT | yes | not established | B | source |
| uniocc | TASL Lab | not established | not established | MIT | not established | no | B | source |
| Validated ground truth for COCO, Pascal VOC, KITTI and Cityscapes (RELiABEL re-annotation) | Osnabrueck University | not established | not established | not established | not established | yes | B | source |
| ZimaBlue pretraining corpus (120K h embodied video) | Joy Future Academy | 120000 | not established | weights not released as of 2026-09-15 (GitHub repo holds README + teaser only) | no | not established | B | source |
Cite as: Blomega Data Refinery, https://data.blomega.com, CC BY 4.0. Published by
Blomega (Wikidata Q141048865). Unknown values are shown as
not established, never guessed.