Datasets:

LightwheelAI
/

EgoDemo

This Data Use Agreement (the "Agreement") governs access to and use of the LightwheelAI datasets covered by Section 1. By accessing, downloading, copying, or using a covered Dataset, the individual or legal entity doing so ("Recipient") agrees to be bound by this Agreement. If an individual accepts this Agreement on behalf of an organization, that individual represents that they have authority to bind the organization.

This Agreement is a custom data license. It is not an OSI-approved open-source license and does not place any Dataset in the public domain. Describing a Dataset as public, a demo, or a trial does not authorize resale, mirroring, or redistribution outside this Agreement.

1. Datasets Covered by This Agreement

"Dataset" means a dataset released by Licensor and expressly made available under this Agreement, together with the materials made available through its repository. The covered dataset family includes the following repositories when they designate this Agreement as their governing data license:

LightwheelAI/EgoStandard — https://huggingface.co/datasets/LightwheelAI/EgoStandard
LightwheelAI/EgoPro — https://huggingface.co/datasets/LightwheelAI/EgoPro
LightwheelAI/EgoDemo — https://huggingface.co/datasets/LightwheelAI/EgoDemo

Other LightwheelAI datasets are covered only if their repository or accompanying release documentation expressly adopts this Agreement. References to the Dataset apply to each covered dataset that Recipient accesses or uses.

Depending on the dataset and variant, the Dataset includes, as applicable:

head-view egocentric video;
synchronized head-view and wrist-view egocentric video;
left- and right-hand pose data;
full-body pose data included in motion-focused variants or samples;
wrist-camera pose or calibration data when supplied;
V4.1 temporal semantic annotations and other annotations released under this Agreement;
raw-video variants and their technical metadata;
task, subtask, scene, episode, quality-control, calibration, source-provenance, and lineage metadata;
manifests, schemas, checksums, documentation, and sample assets; and
any later correction or update that states that it is governed by this Agreement.

EgoStandard includes the EgoStand and EgoStand-body Sub-SKUs. EgoPro includes the EgoProStandard and EgoProStandard-body Sub-SKUs. EgoDemo includes annotated samples from those four variants, together with the head-raw and headwrist-raw trial variants. EgoDemo samples may originate from the formal EgoStandard or EgoPro repositories and retain their source provenance.

Software code, third-party components, and materials expressly accompanied by a separate license are governed by their respective licenses.

"Dataset Materials" means the Dataset and any copy, subset, sample, extract, reformatted version, translated annotation, or other transformation that retains the underlying records or permits reconstruction of all or a substantial portion of those records. Any question concerning the scope of “Dataset Materials” is subject to Licensor’s good-faith determination. Recipient must direct any such question to Licensor before relying on its own reading of the Agreement. Licensor’s written clarification is final and binding on the specific facts presented. Recipient acknowledges that reliance on its own interpretation, rather than seeking clarification, does not excuse a subsequent breach of Section 3.

2. License Grant and Permitted Uses

Subject to continued compliance with this Agreement, Licensor grants Recipient a worldwide, non-exclusive, royalty-free, non-transferable, and non-sublicensable license to:

access, download, reproduce, clean, label, format, combine, and otherwise process Dataset Materials within Recipient's controlled environment;
use Dataset Materials for commercial and non-commercial machine-learning research, training, pre-training, fine-tuning, evaluation, benchmarking, validation, and product development;
develop and commercially deploy models trained or evaluated using the Dataset; and
create and use aggregate statistics, research findings, evaluation results, and other non-reconstructive analytical results.

For EgoDemo, permitted uses also include loading-pipeline integration and visualization.

Recipient represents that it possesses the legal bases, authorizations, policies, and organizational measures required under applicable law to process such data for its intended purposes, and that it will not process Dataset Materials for any purpose not permitted by this Agreement.

Recipient may allow its employees, controlled affiliates, contractors, and cloud service providers to process Dataset Materials solely for Recipient's permitted purposes, provided that each such party is bound in writing by confidentiality, security, privacy, and use restrictions at least as protective as this Agreement. Recipient remains responsible for their conduct and must not permit independent use of the Dataset by those parties.

3. No Sale, Resale, or Redistribution of Dataset Materials

Recipient must not, directly or indirectly:

sell, resell, rent, lease, license, sublicense, assign, publish, distribute, donate, disclose, or otherwise provide Dataset Materials to a third party;
mirror or re-host Dataset Materials, whether publicly or behind a login;
make Dataset Materials available through a marketplace, download service, data exchange, API, hosted query service, or similar offering;
include Dataset Materials in a product or service whose purpose or material effect is to provide access to the original records or substantially equivalent records;
evade these restrictions through superficial changes, format conversion, relabeling, filtering, splitting, augmentation, or combination with other datasets; or
remove or obscure copyright, source provenance, lineage, consent, attribution, or license notices included with the Dataset.

Providing access to controlled service providers under Section 2 is not a prohibited redistribution. For EgoDemo, publishing a list of record identifiers, manifests, or metadata that enables third parties to locate, reconstruct, or retrieve the trial records is prohibited.

4. Trained Models, Model Weights, and Outputs

Recipient may use, distribute, license, sublicense, sell, and commercially deploy a model trained using the Dataset (a "Permitted Model") if the model, its weights, and its ordinary outputs do not contain, reveal, reproduce, or enable extraction or reconstruction of Dataset Materials, personal data, or recognizable source recordings in reasonably foreseeable use.

A model or its weights are not treated as Dataset Materials solely because the Dataset was used for training, fine-tuning, or evaluation. This exception does not apply to a model, embedding collection, retrieval index, output, or other artifact that stores, memorizes, exposes, or functions as a substitute for Dataset Materials.

Recipient must not intentionally train, configure, prompt, or deploy a model for the purpose of reproducing Dataset Materials. Recipient must apply reasonable testing and safeguards against memorization, extraction, and regurgitation of source records.

5. De-Identification, Consent, and Privacy

The Dataset contains human activity data. Licensor states that participants provided informed consent covering data collection, annotation, controlled dataset release, and uses permitted by this Agreement.

Before release, the Dataset underwent a de-identification process designed to remove, obscure, or mask direct personal identifiers identified in the release workflow. "De-identified" does not mean that re-identification is impossible or that the Dataset qualifies as anonymous data under every applicable law. Raw video variants remain subject to all privacy and no-re-identification obligations in this Agreement.

Recipient must not:

attempt to identify or re-identify a participant, bystander, household, employee, facility occupant, or other person represented in the Dataset;
use facial, voice, gait, body, location, biometric, or other signals in the Dataset to determine or verify a person's identity;
link Dataset Materials with external information for identification or re-identification purposes;
contact, target, profile, make decisions about, or publish information about a person based on an actual or suspected identity derived from the Dataset; or
disclose an accidental or suspected identification to any person other than Licensor or a legally required authority.

Recipient must implement reasonable technical and organizational safeguards, limit access to personnel with a need to know, and comply with applicable privacy, data-protection, cybersecurity, employment, biometric, and human-subject laws. Recipient is independently responsible for determining whether its use requires additional notices, approvals, impact assessments, or legal bases. Licensor has no control over, and assumes no responsibility or liability for, Recipient’s processing activities after Dataset Materials are made available to Recipient.

Recipient must promptly notify Licensor at contact@lightwheel.ai without undue delay and in any event no later than seventy-two (72) hours after becoming aware of exposed personal information, a credible re-identification risk, unauthorized access, or a security incident involving Dataset Materials, and must cooperate in reasonable mitigation steps.

6. Attribution and Public Statements

In publications, technical reports, dataset comparisons, and public model documentation where dataset attribution is customary, Recipient must identify each Dataset used by its repository name, such as LightwheelAI/EgoStandard, LightwheelAI/EgoPro, or LightwheelAI/EgoDemo, and include a link to the corresponding repository.

For EgoDemo samples originating from EgoStandard or EgoPro, Recipient must preserve the included source-provenance and lineage information. Recipient must not describe EgoDemo samples as newly collected or as additional hours beyond their source collection totals.

This Agreement does not grant permission to use Licensor's names, trademarks, or logos in a manner that suggests endorsement, sponsorship, or partnership.

7. Ownership and Reservation of Rights

The Dataset is licensed, not sold. Licensor and its applicable rights holders retain all right, title, and interest in the Dataset. No rights are granted except those expressly stated in this Agreement. Recipient owns its independently created Permitted Models and non-reconstructive results, subject to Licensor's rights in the Dataset Materials and any third-party rights.

8. Compliance Records

Recipient must maintain reasonable records sufficient to identify who has access to Dataset Materials and the systems in which they are stored. Upon a credible written allegation of a material breach, Recipient must provide Licensor with a reasonable written description of its compliance measures. This section does not require disclosure of trade secrets, model weights, customer data, or security information beyond what is reasonably necessary to address the allegation.

9. Term and Termination

This Agreement remains in effect while Recipient possesses or uses Dataset Materials. Licensor may terminate the license for a material breach that remains uncured fifteen (15) days after written notice. Licensor may terminate immediately for intentional raw-data resale or redistribution, deliberate re-identification, or a material security violation.

Upon termination, Recipient must stop accessing and using Dataset Materials and, within thirty (30) days, delete all copies under its control, except copies kept solely where required by law or in ordinary disaster-recovery backups that are not restored for active use. Recipient must confirm deletion in writing upon reasonable request.

Permitted Models and non-reconstructive results created while Recipient was in compliance may continue to be used and distributed after deletion of Dataset Materials, provided that they remain compliant with Sections 3 through 6 and do not contain, reveal, or enable reconstruction of Dataset Materials or personal data.

10. Disclaimer of Warranties

TO THE MAXIMUM EXTENT PERMITTED BY LAW, THE DATASET IS PROVIDED "AS IS" AND "AS AVAILABLE," WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, SECURITY, OR LEGAL COMPLIANCE. LICENSOR DOES NOT WARRANT THAT ALL IDENTIFIERS HAVE BEEN DETECTED OR THAT RE-IDENTIFICATION IS IMPOSSIBLE. LICENSOR MAKES NO COMMITMENT REGARDING THE AVAILABILITY, CONTINUITY, OR MAINTENANCE OF THE DATASET. LICENSOR MAY MODIFY, SUSPEND, LIMIT, OR WITHDRAW ANY DATASET OR PART OF IT, INCLUDING HOSTING ON ANY PLATFORM, AT ANY TIME AND FOR ANY REASON, WITH OR WITHOUT NOTICE. NO ACCESS TO DATASET MATERIALS, AND NO CONTINUED AVAILABILITY OF DATASET MATERIALS, IS CONSTRUED AS A CONTRACTUAL, IMPLIED, OR OTHERWISE ARISING OBLIGATION OF LICENSOR. RECIPIENT IS SOLELY RESPONSIBLE FOR MAINTAINING ITS OWN BACKUP COPIES OF DATASET MATERIALS CONSISTENT WITH THIS AGREEMENT.

11. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY LAW, LICENSOR AND ITS AFFILIATES, CONTRIBUTORS, AND DATA PROVIDERS WILL NOT BE LIABLE FOR INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, EXEMPLARY, OR PUNITIVE DAMAGES, OR FOR LOST PROFITS, REVENUE, DATA, OR GOODWILL, ARISING FROM OR RELATED TO THE DATASET OR THIS AGREEMENT. LICENSOR'S TOTAL AGGREGATE LIABILITY WILL NOT EXCEED THE GREATER OF USD 100 OR THE AMOUNT RECIPIENT PAID LICENSOR SPECIFICALLY FOR ACCESS TO THE DATASET DURING THE TWELVE MONTHS BEFORE THE EVENT GIVING RISE TO LIABILITY.

12. Indemnification

To the extent permitted by law, Recipient will defend, indemnify, and hold harmless Licensor and its affiliates, contributors, and data providers from third-party claims, damages, penalties, and reasonable costs arising from Recipient's breach of this Agreement, unlawful use of the Dataset, prohibited redistribution, re-identification activity, or failure to protect Dataset Materials.

13. Governing Law and Dispute Resolution

This Agreement, and any dispute or claim arising out of or in connection with it or its subject matter or formation (including non-contractual disputes or claims), is governed by the laws of the State of California, without regard to its conflict-of-laws rules. The United Nations Convention on Contracts for the International Sale of Goods does not apply.

The parties shall attempt in good faith to resolve any dispute through senior management negotiation within ten (10) business days of written notice. If not resolved within thirty (30) days of initial notice, any dispute shall be finally resolved by binding arbitration administered by JAMS (or AAA, as Licensor elects) under its commercial arbitration rules, conducted in Santa Clara, California, in English. The award shall be final and may be entered in any court of competent jurisdiction.

Notwithstanding the foregoing, either party may apply to a court of competent jurisdiction for urgent interim or injunctive relief, including to prevent actual or threatened unauthorized distribution, re-identification, or security compromise involving Dataset Materials. Recipient acknowledges that any unauthorized use, reproduction, or distribution of Dataset Materials, or any unauthorized disclosure of personal data contained in Dataset Materials, may cause irreparable harm to Licensor and its rights holders for which monetary damages, including the limitation in Section 11, would be an inadequate remedy, and that Licensor is therefore entitled to seek injunctive and equitable relief, without prejudice to all other rights and remedies available at law or in equity.

14. General Terms

Recipient must comply with applicable export-control, sanctions, anti-corruption, and other laws. Recipient represents and warrants that it is not, and is not owned or controlled by, and is not acting on behalf of, any person that is (a) located in, organized in, or a resident of any country or territory subject to comprehensive sanctions administered by the United States, the European Union, the United Kingdom, or the United Nations; or (b) identified on any applicable restricted-party or sanctions list. Recipient must not access, download, or use Dataset Materials from any such jurisdiction or for the benefit of any such person.

Recipient may not assign this Agreement without Licensor's prior written consent, except in connection with a merger or sale of substantially all of Recipient's relevant business where the successor agrees in writing to this Agreement.

If any provision is unenforceable, it will be modified only to the minimum extent necessary, and the remaining provisions will remain effective. A waiver must be in writing and applies only to the specific instance stated. This Agreement is the complete agreement concerning the Dataset unless Recipient and Licensor sign a separate written agreement that expressly supersedes it.

Sections 3 through 15 survive termination to the extent necessary to give them effect.

This Agreement is drafted in the English language. English is the controlling language for all purposes. Any translation is provided for convenience only and has no legal effect.

15. Effective Date, Versioning, and Amendments

This Agreement takes effect, with respect to a given Recipient, on the date the Recipient first accesses, downloads, or uses Dataset Materials (the “Effective Date”), and continues in effect as provided in Section 9.

Licensor may publish amended or successor versions of this Agreement. Any amended version applies only to Dataset Materials first accessed, downloaded, or released after the amended version takes effect, unless Recipient expressly accepts the amended version in writing, in which case it applies to Recipient’s continued use of Dataset Materials. Dataset Materials already obtained by Recipient remain governed by the version of this Agreement in force when they were obtained, except where a later version is adopted by express agreement.

Licensor may correct clerical errors or update references (including repository locations, dataset names, and contact information) without such corrections constituting an amendment under this Section.

16. Contact

Licensing questions: contact@lightwheel.ai
Privacy or security reports: contact@lightwheel.ai