PowerMind VisionID Workflows
Introduction
PowerMind VisionID provides a browser-based workflow for authorized users to sign in, enroll a face, identify a person from a photo or live camera capture, and verify a person against a known ID.
Open VisionID from Tools in PowerMind Studio. Your organization may also provide a direct VisionID URL.
Feature Overview
- sign in with username and password
- enroll a person from a JPG or PNG image
- optionally capture a live face photo from the browser camera for enrollment
- identify a face from a single photo
- identify a live person from a camera frame sequence
- verify a face against a specific enrolled
person_id - inspect match confidence and liveness indicators
Roles
superadmin: can manage users and biometric workflows.admin: can manage lower-privilege users and biometric workflows subject to backend role rules.viewer: can identify and verify faces, but cannot enroll people or manage users.
Face Enrollment Workflow
- Sign in to the UI.
- Open the
ENROLLtab. - Enter a unique
person_id. - Enter the person’s full name.
- Choose a JPG or PNG file, or start the enrollment camera and use a live capture.
- Optionally enable overwrite if the
person_idalready exists. - Submit the enrollment.
Enrollment succeeds only when the server can decode the image and detect exactly one face.
Practical Tips
- Use a frontal, well-lit photo.
- Avoid sunglasses, masks, blur, and multiple people in the frame.
- If you capture from the camera, keep the face centered and still long enough for a good frame.
Face Identification Workflow
- Sign in to the UI.
- Open the
IDENTIFYtab. - Start the camera.
- Look directly at the camera and blink naturally.
- Click
CAPTURE & IDENTIFY.
The browser captures 20 frames at about 5 fps, sends them to the server, and then shows:
- match status
- matched person ID and name when a match is found
- confidence score
- liveness indicators for blink and movement
Liveness Detection Workflow
Live-video identification uses two signals:
- blink detection based on eye aspect ratio
- slight micro-movement of the nose position across frames
If the capture looks like a static photo or replay, the server rejects it with a spoof verdict.
What To Do If Liveness Fails
- blink naturally during capture
- hold the device normally instead of locking it in place
- make sure the camera sees one face clearly in enough frames
- use a live person rather than a still image shown to the camera
Common Usage Scenarios
Enroll a New Employee
- sign in as
superadminoradmin - add the employee with a unique
person_id - verify the stored person name and enrollment timestamp
Identify a Visitor
- sign in as a role that can identify faces
- use the live camera workflow
- review the returned name and confidence score
Verify a Claimed Identity
- send a known
person_idand a face photo to the verification endpoint - review whether the server reports a match
Resolve a Failed Match
- confirm the person was enrolled first
- compare lighting and pose between the enrollment and probe image
- check whether the uploaded face contains a single clear face
Frequently Asked Questions
How do I get access?
Use your organization account. If VisionID is not available in your Studio project, ask your PowerMind administrator to confirm your project membership and role.
Why does the browser need HTTPS?
Most browsers require HTTPS for camera access through getUserMedia() when the site is opened through a host name or IP address.
Why does identification return "No match found"?
The enrolled photo may not be close enough to the probe image, the face may be too different in pose or lighting, or the similarity score may be below the configured threshold.
Why is enrollment rejected?
Common causes are:
- the file is not JPG or PNG
- the image cannot be decoded
- no face is detected
- multiple faces are detected
- the
person_idalready exists and overwrite was not enabled
Can I use the UI without the browser camera?
Yes. Enrollment also accepts photo upload, and verification accepts photo upload.