A guided tour through the SYNAPSIS pipeline — from understanding what video masking is, through the four tools that take a sensitive recording to a citeable, FAIR-archived dataset.
Privacy–utility trade-offs, the five types of identifying information, and when masking is worth your time.
What MaskAnyone does, where it sits in the pipeline, and a 90-second tour of running a job end-to-end.
Pose-accuracy and kinematic metrics — the numbers that let you defend masked outputs to a reviewer or data steward.
Dashboards, batch processing, QA review, archive prep. When the laptop full of progress bars stops working.
Packaging masked outputs with FAIR metadata, choosing an access tier, and depositing into DANS Data Station SSH.
Detailed mechanics that don't fit in the pipeline-tour videos. Each companion goes deep on a specific surface — installation, the hands-on first job, structured quality assessment, and troubleshooting.
Docker, GPU support, starting the stack, recognising a healthy install.
Six-step walkthrough: upload, select people, pick a masking style, run, review, refine.
Structured sampling protocol, privacy vs utility checks, QA paperwork your data steward will accept.
Six common failure modes — install errors, container crashes, OOM, detection failures, tracking loss, GPU missing.
Scripts for the full series are drafted. Recording, editing, and publication are scheduled across 2026–2027 under the SYNAPSIS RA-Training workstream (WP1, batch 1; WP2, batch 2). Source markdown lives at team/ra_training/video-tutorials/scripts/. Open an issue on the SYNAPSIS repository to flag content corrections or request additional topics.