🧑

Face Recognition

Biometric Workshop Suite — Feature Visualisation

← All Modules
📷
Loading face detection models…
Overlay
Loading models from CDN…
Jawline
Eyebrows
Nose
Eyes
Mouth
Detection
Status Idle
Bounding box —
Face size —
Detection score 0%
Facial Landmarks (68 pts)
Jawline 17 pts
Eyebrows 10 pts
Nose 9 pts
Eyes 12 pts
Mouth 20 pts
Extracted Biometric Features
Start camera to see live features
Feature Vector Heatmap (16 features)

Each cell is one geometric feature normalised 0–1. Darker = smaller value. In deep-learning systems, this vector has 128 dimensions.

Face Recognition

Capture 3 times — move naturally between each shot.

Enrolled profiles
No profiles enrolled yet
How face recognition works
📷
1. Capture
A camera frame is grabbed and passed to the detector
→
📦
2. Detect
A neural net (SSD MobileNet) finds face bounding boxes in the image
→
📍
3. Landmarks
68 key points are located on the face (eyes, nose, mouth, jaw)
→
📏
4. Geometry
Distances and ratios between landmarks form a compact feature vector
→
🔍
5. Match
The vector is compared against stored templates using cosine or Euclidean distance