Workflows
6
in the AudioPipeline-style folder
Faces,
detected.
Real-time video on Protor. Webcam face detection, landmark overlay, live privacy-blur, RTSP surveillance, evidence-grade video archive. Same engine, six workflows — choose what your room needs.
Sources
webcam · RTSP · file
live + archive
Vision
AIVision
FaceDetect · DetectionOverlay
Default
privacy
blur-on by design
Real-time vision,
without the cloud.
Most live video AI ships as a SaaS. Frames go up, judgements come down, and the building owner gives up custody of who walked through the lobby at 11:42.
This pipeline does it on the camera or the workstation. Vectorio.Video.Webcam.Subscriber grabs frames. Vectorio.AIVision.FaceDetect runs the detector. Vectorio.AIVision.DetectionOverlay annotates the live feed. Optional Vectorio.Video.PrivacyBlur ships the same frame with faces blurred before anything is recorded.
The same engine drives RTSP surveillance feeds, forensic video archive workflows, and the privacy-blur livestream you'd want by default in a public space.
Four nodes,
one .viow file.
From E2E Tests/Video/webcam-facedetect-overlay.viow. The baseline workflow — extend by adding a publisher, a redactor, a recorder.
ingestvideo.Webcam.Subscriber
→
detectaivision.FaceDetect
→
annotateaivision.DetectionOverlay
→
emitvideo.Viewer
Six workflows,
same engine.
| Workflow | What it does |
|---|---|
webcam-viewer-simple.viow | Bare webcam capture → viewer. The starting point. |
webcam-facedetect-overlay.viow | This page — detect + overlay live boxes on the feed. |
webcam-landmarks-overlay.viow | Face-landmark annotations (eyes, mouth, jaw) overlaid live. |
webcam-face-blur-live.viow | Live privacy: blur faces before they leave the device. |
rtsp-surveillance-analytics.viow | RTSP camera feed → detection + analytics → downstream. |
video-file-evidence-processing.viow | Forensic video file → process → signed evidence pack. |
Why this matters.
- Frames never leave the box. The model file and the camera live in the same process. There is no upload, no API, no third-party access to who walked in.
- Privacy by default. The blur workflow is one node away from any other. Public-space cameras can be deployed with blur-on and still produce evidence-grade archives.
- Same shape as the rest. Video pipelines pack into the same forensic evidence format as mail archives and NIS 2 findings. One verifier, three domains.