Privacy and consent

CERES records human-facing research data. Plan privacy and consent before enabling the outward camera or microphone. The study team is responsible for following its approved protocol, institutional requirements and data-governance rules.

What a capture can contain

Outward video and microphone audio can identify people, conversations, rooms, screens and documents. CERES also records head pose, hand joints, task labels, task outcomes and run metadata. Explain what the study records, why it is collected, where it will be kept, who may access it and how long it will be retained before the run begins.

The run editor can optionally attach a study-assigned headset ID, demonstrator ID, project ID, consent date and consent-document ID to the recording. Use study identifiers rather than a participant's name. When the demonstrator ID is blank, CERES generates a readable designator for that episode. It is not a substitute for the study's consent record.

Consent the observed person

The person performing the task may be recognisable in the outward camera view or microphone audio even when they are not the person wearing the headset. Obtain and document their informed consent in the form required by the study before recording. Give them a clear chance to ask questions, decline, pause or stop their involvement without pressure.

Agree a simple stop or pause signal before the headset goes on. The capture director and demonstrator should both know who can use it and how the run will be marked if it is interrupted.

Include other participants deliberately

Anyone who may appear in video, be heard in audio or contribute to the task is a participant in the capture boundary. Tell them what will be recorded and obtain the consent required by the study before they enter that boundary. This includes assistants, interviewers and people handling task objects.

Avoid recording in public or uncontrolled spaces. If a visitor or bystander enters the camera or microphone area unexpectedly, pause recording and resume only when the agreed conditions are restored. Do not treat incidental footage or audio as consent.

Prepare a privacy-safe capture space

Before arming the recorder:

1. Check that the camera view contains only people who are allowed to be recorded.
2. Remove or cover names, paperwork, screens, badges and other sensitive material that are not part of the study.
3. Decide whether microphone audio is required. Disable it when the protocol does not need it.
4. Confirm the approved storage location, access group and retention period for the run.
5. Record only the task needed for the study. Pause during setup conversations, reset periods and unrelated activity.

During and after the run

Do not start the important motion until CERES shows REC. If any participant asks to pause or stop, use the run controls before continuing. Record the resulting decision and follow the study protocol for an interrupted run or a request to withdraw.

Keep browser storage, local folders and exported datasets within the study's access controls. Treat a Hugging Face sync as a deliberate export of the selected dataset, not as a substitute for the study's data-management approval. Do not share a recording, its viewer link or its exported artefacts outside the approved group.

When raw microphone audio is disabled, CERES does not retain audio chunks in the recording. In paired capture, audio sent to speech recognition follows a separate non-durable path. Consent still needs to cover that use.

For camera permissions, storage and a first-run preflight, read Setup and system requirements. For the data that an export contains, read Data description and output format.