Give each review unit an identity
Choose identifiers for sessions, derived clips, and task episodes. Explain whether a file contains one task or several and how repeated attempts are represented. Keep identifiers stable when filenames change. A manifest should let a buyer trace an annotation back to the actual recording, rather than relying on a directory name that may be edited during delivery.
Describe useful capture context
Include the camera configuration, viewpoint, task labels, and relevant environment descriptors that the project needs. Distinguish observed fields from information inferred during review. If values are missing, explain whether they were not collected, withheld, or not applicable. Consistent missing-value meaning is more useful than filling every row with a vague default.
Link original and derived assets
Record source-session intervals for clips and references for annotations, transcripts, or optional sensor data. Include file integrity information where agreed and identify the delivered version. A later crop or redaction should have an explicit relationship to its source so reviewers know whether existing spatial labels and timestamps still apply to the changed media.
Keep status separate from content
Technical review, annotation review, and proposed permission scope are different kinds of information. Do not collapse them into one unexplained approved field. Define what each status means and who maintains it. A source record can help organize review, but it is not itself proof that every intended use has been cleared or that the collection satisfies the buyer’s task.
Review checklist
- Identify sessions and derived clips
- Define fields and missing values
- Link annotations to media versions
- Explain each review status