Product Specification

Swing Gate Turnstile Face Recognition 1400mm 304 Stainless Steel Acrylic Arm Liveness Detection

Product Introduction

A biometric lane only works if people trust it: it has to open on the first look, in bad light, with a mask on, and it has to be obvious that a photo held up to the camera will not get anyone in. This swing gate turnstile pairs a face recognition terminal with the lane controller, recognises a registered person in about 0.3 seconds against a gallery of up to 30,000 faces, and applies liveness detection to reject a printed photo, a phone video or a mask. Near infrared fill light handles a backlit lobby entrance where the glass wall sits behind the visitor, and the camera still matches a person wearing glasses, a cap or a medical mask by using the visible eye and forehead region. Templates are stored as encrypted feature vectors, not as photographs, so the system can be run without keeping images of employees and visitors, which is what a GDPR or a local data protection review usually asks first. When a face is refused, the lane falls back to an IC card, an app QR code or an NFC phone without a second queue, and every attempt is logged with the result. The 1400mm stainless steel cabinet with a 550-1000mm channel opens in 0.5 seconds, passes 30 people per minute, 40 in normally open mode, and keeps a local template cache so the lane still works when the server or the network is down.

Product Features & Advantages

1. About 0.3 Seconds, Up To 30,000 Faces

A registered person is matched in roughly 0.3 seconds against a gallery of up to 30,000 faces held on the terminal and the controller. A 1:N search at that size covers a tower, a campus or a factory without splitting people across readers.

2. Liveness Detection Against Photos And Video

The camera runs a liveness check on texture, depth and micro movement before it accepts a match, so a printed photo, a phone screen replay or a silicone mask is refused even when the face itself is genuine. That is the control an audit actually tests.

3. Backlight, Low Light And Evening Entrances

Near infrared fill light and wide dynamic range handle a glass wall behind the visitor and a dim service entrance at night. The lane reads the face before the person stops walking, which is what keeps a morning peak moving.

4. Glasses, Caps And Medical Masks

The match uses the eye and forehead region, so glasses, a cap or a mask worn for health reasons does not force a person to remove it at the gate. Staff who prefer not to enrol a face keep using a card, and both methods coexist on the same lane.

5. Templates Encrypted, No Photo Archive

Only an encrypted feature template is stored, not a photograph, and a person's template can be deleted when they leave. This keeps a biometric deployment inside a privacy review instead of outside it, and the record shows a match result rather than an image.

6. Graceful Fallback And One Record Per Passage

A refused or failed face falls back to an IC card, an app QR code or an NFC phone on the same lane, with no second queue and no staff intervention. Every attempt, accepted or refused, is logged with time, lane and method, and four pairs of infrared sensors confirm that one person passes per accepted match.

Frequently Asked Questions

1. Can a photo or a video on a phone open the lane?

No. Liveness detection checks texture, depth and micro movement, so a printed photo, a replay on a phone screen or a silicone mask is refused even when the face belongs to a registered person. The refusal is logged as a spoof attempt.

2. How many faces can be enrolled and how fast is the match?

The terminal and controller hold up to 30,000 templates with a match in about 0.3 seconds at 1:N. Larger sites are split by lane group, so each lane only searches the population that uses it.

3. Do glasses, caps or masks stop the recognition?

Glasses, a cap and a medical mask are tolerated because the match uses the eye and forehead region. A person who prefers not to be recognised by face keeps using an IC card or a phone credential on the same lane.

4. Are face photos stored, and does this meet GDPR style rules?

Only an encrypted feature template is stored, not a photograph, and the template can be deleted when a person leaves. The record stores the match result, the time and the lane, which is the usual basis for a data protection review.

5. What about lookalikes, and does the lane work offline?

The acceptance threshold is tunable, and a close match can be set to require a second factor such as a card or a PIN. Templates are cached in the local controller, so recognition keeps working during a server or network outage and the records upload on reconnect.