Product Specification

Wing Barrier Turnstile Face Recognition Liveness Anti Spoof 0.3s Identification 50000 Face Library

Product Introduction

This wing barrier turnstile is fitted with a dual lens face recognition terminal that identifies a person in 0.3 seconds and opens the acrylic wings immediately, so staff and members walk through without stopping, taking out a card or touching any surface. The terminal holds up to 50,000 face records locally, which is enough for a large campus, a factory site or a residential community to run the whole entrance on faces alone.

Liveness detection is the core of the design. The near infrared lens checks depth and skin reflection while the visible light camera compares the facial template, so a printed photo, a phone screen or a video replay is rejected at the lane. Recognition also works with a mask or a cap covering part of the face, and the infrared fill light keeps accuracy in dim corridors and at night.

Enrolment is done from the management platform or directly on the terminal: a face is captured in a few seconds and bound to a person record, an employee number or a department. Faces can be grouped by company, floor or class, so one lane serves several organisations while each group sees only its own people. The terminal exposes an SDK and HTTP interface for integration with HR, attendance and visitor systems, and the lane reports each recognition result with person ID, lane, direction and timestamp.

Product Advantages

• 0.3 second identification - The dual lens terminal returns a recognition result in about 0.3 seconds and the wings open at once, so a person walks through at normal pace without pausing in front of the camera.
• Liveness detection against photo and video spoof - Near infrared depth checking combined with visible light template matching rejects printed photos, phone screens and recorded video, closing the most common way of bypassing a face terminal.
• Mask and partial face recognition - Recognition stays reliable when a mask, cap or glasses cover part of the face, and the infrared fill light maintains accuracy in dim corridors, basements and at night.
• 50,000 face records on the terminal - The local face library holds up to 50,000 people with a matching accuracy of 99.5 percent or better, so a campus or industrial site can run the entrance without a server round trip for every person.
• Group based permission - Faces are organised by company, floor, class or shift, and each group can be limited to specific lanes and time windows, allowing one lane to serve several organisations with separate data.
• SDK and HTTP integration - The terminal provides an SDK and HTTP interface for connection to HR, attendance, visitor management and payroll platforms, and recognition records carry person ID, lane, direction and timestamp.
• Anti tailgating and anti crush - Six pairs of infrared sensors detect a second person following a recognised user and keep the wings closed, while the anti crush mechanism stops the arms on contact.
• Automatic reset and emergency release - An unused permission is cancelled after 10 seconds to prevent misuse, and on power failure or emergency stop the wings unlock so the lane never blocks evacuation.

FAQ

Q1: How fast is face recognition at the lane?

A recognition result is returned in about 0.3 seconds and the wings open immediately, so a person passes without breaking stride. The lane itself handles 30 to 40 people per minute.

Q2: Can a photo or a phone screen fool the terminal?

No. The terminal uses near infrared depth checking together with visible light template matching, so printed photos, phone screens and video replays are rejected as non living faces.

Q3: Does it recognise people wearing masks?

Yes. Recognition remains reliable with a mask, cap or glasses, and the infrared fill light keeps accuracy in dim corridors and at night.

Q4: How many faces can be stored?

Up to 50,000 face records are stored on the terminal with a matching accuracy of 99.5 percent or better, so the entrance can run without contacting a server for each person.

Q5: Can several companies share one lane?

Yes. Faces are grouped by company, floor, class or shift with separate lane and time permissions, so one lane can serve multiple organisations while keeping the data apart.

Q6: Can it export attendance data?

Yes. Each recognition record carries person ID, lane, direction and timestamp, and the terminal exposes an SDK and HTTP interface for HR, attendance and payroll systems.

Q7: What happens in a power failure or an emergency?

The wings unlock automatically on power failure, and the emergency stop button releases every lane at once, keeping the exit route clear.