The Robotics Division at vteams has developed a real-time Automatic Face Recognition Application using core Python, LBPH algorithm and OpenCV 2.4.10 library. Facial recognition being the most convenient biometric technology, works with the most apparent individual identifier – the human face.
In an experiment, we used Face Recognition System to monitor employee attendance. We programed this application to automatically detect and recognize when particular employees enter or leave the office premises.
Face Recognition Application takes into account characteristics of a person’s facial features fed by a high resolution digital video camera. The application then sends a detected face image to LBPH (Local Binary Patters Histogram) recognition algorithm that starts processing it. It measures the overall features of the face, including distances between nose, eyes, jaw edges, and mouth. These measurements are then stored in a database and used for evaluation when a person stands before the camera.
With every human face having almost 80 nodal points, each face is different from each other by certain landmarks that make up the facial features. Following nodal points are used by our application to identify a human face:
- Distance between the eyes
- Width of the nose
- Depth of the eye sockets
- The shape of the cheekbones
- The length of the jaw line
The following four-stage process explains the way our Face Recognition Application operates:
- Capture – a physical or behavioral sample is captured by the system during registration
- Extraction – a template is created from the unique data extracted
- Comparison – a new sample is then compared with the template
- Matching – the system then decides if the features extracted from the new sample are matching with the training samples or not
A Face Recognition Application for High Traffic Areas:
Our application is also perfect for areas such as:
- Government offices
- Airports and railway stations
- Public transportation
- Business of all kinds
- Corporate sector
On average, our system makes the decision within 5 seconds, which is pretty good when installed at security checkpoints where the traffic is controlled and guided through a set path.