Media
2 items
Description
Inspect physical AI dataset pages for documented rights, provenance, viewpoints, modalities, annotations, and delivery details. Dataset X-Ray reveals what a physical AI dataset page documents—and what it leaves unanswered. Open a dataset page on Hugging Face, GitHub, arXiv, or another website. Click Dataset X-Ray, review the local-analysis disclosure, and choose “Inspect this page.” The extension builds an evidence matrix across 12 procurement and training-readiness areas: - Licence declaration - Commercial-use terms - Redistribution terms - Contributor consent - Privacy and redaction - Data provenance - Viewpoint and capture geometry - Modalities and sensors - Task and action coverage - Annotations and schema - Format and loader - Version and release identity Each documented or ambiguous result includes the exact page text that triggered it. Missing evidence remains clearly marked as not found. Use the result to: - Review a dataset without juggling tabs - Compare documentation with your model and procurement requirements - Export a Markdown evidence report - Turn unresolved areas into a copyable collection-gap brief - Scope missing real-world behaviour data with EGXO Data Privacy by design: - Dataset X-Ray runs only after a user click - It receives temporary access only to the active tab - Page URLs, titles, text, and results are processed locally in Chrome - No page data or results are stored, logged, sold, shared, or transmitted - No accounts, analytics, advertising, tracking, or remote code Dataset X-Ray reports documentation evidence found on one page. It does not provide legal advice, verify that a claim is true, or determine whether a dataset is suitable for a specific project. Always review the source documents and obtain professional advice where appropriate.