01
Why the upload step changes the privacy equation
Cloud redaction can be convenient, but an upload creates another copy, another processor and another set of retention questions. For footage involving employees, claimants, customers or bystanders, that transfer may need its own approval before any redaction work begins.
A local workflow reduces that surface area. The source, detections, corrections and export remain on the operator’s computer. Internet access can be limited to licence validation rather than video transfer.
- Map where the original and exports are stored.
- Confirm that temporary files stay local.
- Record who performs the final visual review.
02
A defensible five-stage workflow
Start with the hardest representative clip, not the cleanest. Import it, select the appropriate privacy mode, allow automatic detection and tracking to complete, then inspect the proposed face tracks before export.
Automation should reduce searching and keyframing; it should not remove human judgement. Review scene cuts, partial profiles, reflections, screens and moments where people enter or leave frame.
- Import locally.
- Detect and track.
- Review suggested gaps.
- Correct misses or false detections.
- Export and spot-check the final file.
03
What to document
For organisational use, keep a short job record: source identifier, operator, selected redaction style, review date and output location. Avoid storing unnecessary personal information in the record itself.
CensorFlow’s privacy review report is designed to support this hand-off by summarising face tracks and moments that deserve attention. It is an operational aid, not a legal conclusion.
Apply the workflow
Review real footage locally.
CensorFlow detects and tracks visible faces on your Windows PC, then lets a person correct and approve the result before export.