RevFace15
User Manual
P a g e
| 52
Copyright©2023 ZKTECO CO., LTD. All rights reserved.
1:1 Threshold Value:
Under 1:1 verification mode, the verification will only
be successful when the similarity between the acquired facial image and the
user’s facial templates enrolled in the device is greater than the set value.
The valid value ranges from 0 to 100. The higher the thresholds, the lower the
misjudgement rate and the higher the rejection rate, and vice versa. It is
recommended to set the default value of 88.
Face Enrollment Threshold:
During face enrollment, 1:N comparison is used
to determine whether the user has already registered before.
When the similarity between the acquired facial image and all registered
facial templates is greater than the set threshold, it indicates that the face has
already been registered.
1: N Match Threshold for Masked People:
The recognition rate of mask
wearing under the setting of 1:N verification mode. The higher the thresholds,
the lower are the misjudgement rate, and higher is the rejection rate, and vice
versa. It is recommended to set the default value of 68.
Face Pitch Angle:
It is the pitch angle tolerance of a face for facial template
registration and comparison.
If a face’s pitch angle exceeds the set value, it will be filtered by the algorithm,
i.e., ignored by the terminal thus no registration and comparison interface will
be triggered.
Face Rotation Angle:
It is the rotation angle tolerance of a face for facial
template registration and comparison.
If a face’s rotation angle exceeds the set value, it will be filtered by the
algorithm, i.e., ignored by the terminal thus no registration and comparison
interface will be triggered.
Image Quality:
It is the image quality for facial registration and comparison.
The higher the value, the clearer image is required.
Minimum Face Size:
It sets the minimum face size required for facial
registration and comparison.
If the minimum size of the captured image is smaller than the set value, then
it will be filtered off and not recognized as a face.
This value can also be interpreted as the face comparison distance. The
farther the individual is, the smaller the face, and the smaller number of pixels
of the face obtained by the algorithm. Therefore, adjusting this parameter can
adjust the farthest comparison distance of faces. When the value is 0, the face
comparison distance is not limited.
