58
SLN-VIZNAS-IOT User Guide, Rev. 1.1, 10/2020
NXP Semiconductors
4 Troubleshooting
4.1 Registering a Face
This section will describe steps that can be taken to help debug potential points of failure
when attempting to perform face registration and recognition. If after following these steps
you still cannot identify the cause of the issue, head over to the
reach out for additional assistance.
4.1.1 Turn off Liveness Detection
First, to help identify whether the issue is with the detection/recognition or with the
liveness detection/anti-spoofing, turn off liveness detection using the
“liveness off”
CLI
) to verify whether or not your face can be
registered without the secure features enabled.
If turning liveness detection off allows your face to able to be registered without any
issues, then the problem is likely related to the liveness detection and anti-spoofing
feature.
Reenable
liveness detection
and skip to
for
steps help identify what specifically is causing liveness detection to fail.
If turning liveness detection off still does not allow your face to be able to be registered,
follow the steps in
to help determine why detection/recognition is
failing.
4.1.2 Change App Type
As discussed in
the E-Lock app type makes use of the IR camera
for detection/recognition while the Door Access app type makes use of the RGB camera
for detection/recognition. To help determine if there is an issue with one specific camera,
try changing from E-Lock to Door Access
or vice versa depending on which configuration
you are currently using.
If after switching to the alternate app type, you are still experiencing an issue, proceed to
the next step.
4.1.3 Debug Using Verbose Mode
Verbose mode can provide useful information when debugging registration issues,
especially when it comes to debugging issues with liveness detection and anti-spoofing.
Follow the steps in the
section of this guide
and enable “verbose
3” to receive all available debugging information. The on-screen indicators can be
particularly helpful in identifying what is going wrong during registration.
For example, in the below screenshots we can see that the IR camera was falsely
detecting a fake face. This is likely due to a lighting issue which is covered in the next
section
