PARAMETERS¶
This section covers the valid values for setting up additional configurations pertaining to face factor.
- class cryptonets_python_sdk.settings.configuration.PARAMETERS(*values)¶
PARAMETERS contains the valid key values for configuring additional parameters in the factor processor.
Configurable Parameter List and Valid values¶
Name |
Description |
Valid Values |
|---|---|---|
K |
Defines the maximum number of nearest neighbors to return for a prediction. If the number of reasonable matches is less than the requested k, only the available reasonable matches are returned. When K=1 or not specified, a single EnrollPredictResult is returned instead of a list of EnrollPredictResult object. |
1 to 100 Default Value is 1 |
INPUT_IMAGE_FORMAT |
Input image format and byte arrangement. |
“rgb”, “rgba”, “bgr” |
CONTEXT_STRING |
Context for computation call. Not applicable for session. |
“enroll”, “predict” (Only allowed for each operation) |
CONF_FAST_PROCESS |
Enable fast search mode. Typically enabled for realtime camera processing. |
True, False |
INPUT_TYPE |
Input type of the image sent (Under Development) |
“face”, “document-id”, “document-barcode” |
FACE_THRESHOLDS_REM_BAD_EMB |
Threshold for geometric distance based embedding removal. |
0, 1, 2 |
BLUR_THRESHOLD_ENROLL_PRED |
Threshold for blur based image rejection for face. Checked on cropped image. |
0 to 10000 |
THRESHOLD_PROFILE_ENROLL |
Threshold for rejecting a (horizontal) profile image - enroll context |
-0.1 to 2 |
THRESHOLD_PROFILE_PREDICT |
Threshold for rejecting a (horizontal) profile image - predict context |
-0.1 to 2 |
THRESHOLD_VERTICAL_ENROLL |
Threshold for rejecting a vertical profile image - enroll context |
-0.1 to 2 |
THRESHOLD_VERTICAL_PREDICT |
Threshold for rejecting a vertical profile image - predict context |
-0.1 to 2 |
THRESHOLD_USER_RIGHT |
Threshold to reject a face if the user face is more aligned sideways - right |
-0.1 to 2 |
THRESHOLD_USER_LEFT |
Threshold to reject a face if the user face is more aligned sideways - left |
-0.1 to 2 |
THRESHOLD_USER_TOO_FAR |
Threshold to reject a face if the user face is too far |
-0.1 to 2 |
THRESHOLD_USER_TOO_CLOSE |
Threshold to reject a face if the user face is too close |
-0.1 to 2 |
IMAGE_BORDER |
Border ( as % of size) to be set while padding. Typically needed for document id |
0 to 0.1 |
IMAGE_PRE_PROC |
Enable additional image processing prior to validity check / cropping |
“zoom_pan”, “rotate90”, “rotate180”, “rotate270”, “blur”, “fliplr”, “none” |
THRESHOLD_GLASS |
Threshold to detect glass (eye wear) |
-0.1 to 2 |
THRESHOLD_MASK |
Threshold to detect mask on a face |
-0.1 to 2 |
ENROLL_ALLOW_EYE_GLASS |
Allow eyeglasses for enroll if within threshold |
True, False |
CONF_SCORE_THR_ENROLL |
Threshold to reject a face if the crop confidence score is lesser than this value - enroll |
-0.1 to 2 |
CONF_SCORE_THR_PREDICT |
Threshold to reject a face if the crop confidence score is lesser than this value - predict |
-0.1 to 2 |
BLUR_THRESHOLD_DOC_LEVEL_1 |
Threshold for blur based image rejection for document type - 1st level checking. full document |
0 to 10000 |
BLUR_THRESHOLD_DOC_LEVEL_2 |
Threshold for blur based image rejection for document type - 2nd level checking. extracted face. |
0 to 10000 |
DOCUMENT_FACE_CHECK_VALIDITY |
Enable cropped face validity check as part of document id verification |
True, False |
DOCUMENT_CHECK_VALIDITY |
Enable cropped document validity check as part of document id verification |
True, False |
DOCUMENT_FACE_PREDICT |
Enable prediction using cropped face image as part of document id verification |
True, False |
MIN_DOCUMENT_BORDER |
Minimum border (in pixels) to be added when padding. typically used for document along with image border |
0 to 10% of image size in pixel |
FACE_DETECT_PREFERRED_SIZE |
Preferred size of the cropped face image. typically used during face extraction from image with multiple faces |
448 |
SEND_ORIGINAL_IMAGES |
Get the original images from the server. Default false |
True, False |
COMPARE_RESERVATION_CALLS |
Number of calls to be set for compare in billing reservation. |
0 to 100000000 |
ESTIMATE_AGE_RESERVATION_CALLS |
Number of calls to be set for estimate age in billing reservation. |
0 to 100000000 |
FACE_ISO_RESERVATION_CALLS |
Number of calls to be set for face ISO in billing reservation. |
0 to 100000000 |
ESTIMATE_AGE_FACE_VALIDATIONS_OFF |
False` by default, If set True, disables the face validation in the age estimation method, the method will return an age estimation unless no face is detected. |
True, False |
COLLECTION_NAME |
Collection name to be used for the operation: possible values are default, RES100``and `RES200. Each collection is tied a sepcfic enmbeddings, you can’t predict a face that was enrolled with different collection, so you need to specify the name of the collection you are targetting. if you leave it empty, the default` collection will be used. |
default, RES100``and `RES200. |
RELAX_FACE_VALIDATION |
False by default, if set to True, the face validation step in any operation is going to be permissive. |
True, False |