Troubleshooting

Face Not Recognised

Practical fixes for when the kiosk does not recognise a worker's face, from re-enrolling with AdaVision ID to adjusting thresholds.

6 min read

Why Does This Happen?

The most common reasons the kiosk fails to recognise a face.

Facial recognition on the kiosk works by comparing the live camera image against the stored face data for each worker. When the system cannot find a confident match, it means the live image looks too different from the stored version. This is usually caused by environmental factors rather than a problem with the system itself.

Poor lighting

The face is in shadow, backlit, or the room is too dark

Wrong angle

The worker is looking away from the camera or standing too far

Appearance change

Sunglasses, hat, or significant change since registration

First-time use

The worker has not registered their face yet

Low-quality original

The original registration was done in poor conditions

Fix 1: Improve the Lighting

Good lighting is the single most important factor.

Lighting makes or breaks facial recognition. The system needs to see the worker's face clearly, with even lighting across the entire face. Here is what to aim for:

Light the face from the front — overhead fluorescent lights or a desk lamp pointed at the kiosk area works well
Avoid backlighting — do not place the kiosk in front of a window or bright light source, as this puts the worker's face in shadow
Consistent lighting — if the lighting changes throughout the day (e.g. natural light), consider adding a fixed light source near the kiosk
Avoid harsh shadows — a single bright spotlight from one side can create strong shadows. Softer, even lighting is better
Lighting Examples
Backlit

Light behind = face in shadow

Front-lit

Light in front = clear face

Fix 2: Face the Camera Directly

The camera needs to see the full face, not a side angle.

For the best recognition, workers should look directly at the camera with their whole face visible. If they are turned to one side, looking down, or standing at a sharp angle, the system may struggle to match their face.

Face the camera straight on — look directly at the lens, not to the side
Stand at the right distance — about arm's length away from the screen
Keep your head level — do not tilt or duck, keep your face centred in the frame
Wait for the green box — a green bounding box appears when the system has detected your face
Tip

If workers are rushing past the kiosk, remind them to pause for a moment and look at the camera. Recognition only takes a second or two, but the camera needs a clear, still image.

Fix 3: Remove Sunglasses and Hats

Some accessories can interfere with recognition.

Accessories that cover significant parts of the face can make recognition difficult. Here is what to keep in mind:

Sunglasses

Remove them — dark lenses hide the eyes, which are a key recognition area

Hats and hoods

Remove or push back — they can shadow the face and hide the forehead

Regular glasses

Usually fine — the system handles normal prescription glasses well

Face masks

Remove briefly for recognition — they cover too much of the face

High-vis vests

No issue — clothing does not affect face recognition

Hard hats

Push back slightly so the face is fully visible

Note

If a worker registered their face while wearing glasses, they should ideally wear glasses when clocking in too (and vice versa). Consistency helps the system match accurately.

Fix 4: Re-Register the Worker's Face

Start fresh with a new face registration.

If a worker's face was originally registered in poor conditions (bad lighting, wrong angle, or wearing accessories), re-registering often solves persistent recognition problems. This clears the old face data and captures a fresh set in better conditions.

Open the worker's profile

In the manage portal, go to Workers and click on the worker who is having trouble.

Clear their face data

In the worker's profile, find the Face Registration section and click Clear Face Data. This removes the stored face descriptor.

Worker re-registers at the kiosk

The next time the worker visits the kiosk, they will be prompted to register their face again, just like their first day. Make sure the lighting is good and they face the camera directly.

Verify recognition

After re-registering, have the worker step back and try clocking in to confirm the new registration works well.

Good to know

When re-registering, make sure the conditions are ideal: good front lighting, face the camera directly, no sunglasses or hats. A high-quality registration makes all future clock-ins smoother.

Fix 5: Use PIN Fallback

An alternative method if face recognition keeps failing.

If a worker consistently has trouble with facial recognition (for example, due to working conditions that make it impractical), they can use a PIN code to clock in instead. PIN entry is always available as a fallback on the kiosk.

Kiosk — PIN Fallback

Enter your Employee ID and PIN to clock in

Employee ID ••••
Clock In

Try face recognition instead

Tip

Workers can switch to PIN entry by tapping the "Use PIN" link on the kiosk screen. They need their Employee ID and PIN, which can be set up in their worker profile.

Understanding Confidence Scores

What the percentage means and how to improve it.

Every time the kiosk matches a face, it assigns a confidence score. This score tells you how certain the system is about the match. Low scores usually indicate an environmental issue that can be fixed.

85%+
High

Strong match. Good conditions. No action needed.

60–84%
Medium

Probable match. Check lighting and angle. Consider re-registering.

Below 60%
Low

Weak match. Re-register the face in good conditions. Try PIN as a backup.

Important

If a worker consistently scores below 70%, clear their face data and have them re-register in ideal conditions (good lighting, facing the camera, no accessories).

Best Setup Tips for Reliable Recognition

Get the kiosk positioned for the best results.

Mount at face height — the camera should be at roughly eye level for the average worker. If it is too high or too low, angles become a problem.
Add a fixed light — place a desk lamp or LED light near the kiosk to provide consistent front lighting, especially in areas with changing natural light.
Avoid direct sunlight — do not place the kiosk where sunlight can hit the camera lens or create glare on the screen.
Keep the lens clean — wipe the camera lens weekly with a soft cloth. Dust and fingerprints reduce image quality.
Mark a standing spot — put a small marker on the floor about arm's length from the kiosk so workers know where to stand.
Register in the same spot — have workers register their face at the same kiosk and in the same lighting they will use daily.
Good to know

A well-positioned kiosk with good lighting can achieve 90%+ confidence scores consistently. Spending a few minutes getting the setup right will save your team time every day.

Ready to try TempClock?

10-day free trial. All features included. No hidden fees, cancel anytime.