External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

Account ID for new account creation.

Account ID for new account creation.

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

Account ID.

An account is a limitation for all the user actions (matching, removing, updating and other) on data which does not have corresponding account ID. For example: it is impossible to attach a face with one account ID to a list with another account ID.

User can make a limited set of requests without this header (requests with methods GET, HEAD, OPTIONS, match requests and getting events statistics requests).

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

Account ID.

An account is a limitation for all the user actions (matching, removing, updating and other) on data which does not have corresponding account ID. For example: it is impossible to attach a face with one account ID to a list with another account ID.

User can make a limited set of requests without this header (requests with methods GET, HEAD, OPTIONS, match requests and getting events statistics requests).

Page number.

Number of items on page.

Account ID filter. Will not have effect if Luna-Account-Id header is used for authorization

Upper bound for object create_time.

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

Account ID.

An account is a limitation for all the user actions (matching, removing, updating and other) on data which does not have corresponding account ID. For example: it is impossible to attach a face with one account ID to a list with another account ID.

User can make a limited set of requests without this header (requests with methods GET, HEAD, OPTIONS, match requests and getting events statistics requests).

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.

Account email.

Type of input image. 0 - raw image, 1 - face warped image, 2 - body warped image.

Whether to estimate people count on the image.

⚠ People counter licensing feature is required to be enabled to perform estimation.

Multiple face detection policy:

• 0 - do not process images with several detected faces.

  If detect_body is enabled, the system will attempt to detect a human body related to the detected face. If a human body is detected, its detection will be associated with the same event as the face.

  If there is no face detected in the image, no human body detections will be returned. Thus there will be no human body detection returned if detect_face is disabled.

• 1 - process images with several detected faces and/or human bodies.

  If detect_body is enabled, all the detections of human bodies will be returned. Face detection and human body detection of the same person will be associated with the same event.

  The detections of human bodies are returned even if there were no faces found.

• 2 - get the best face detection from the image. A single face of the best quality is detected in the image.

  If detect_body is enabled, the system will attempt to detect a human body related to the detected face. If a human body is detected, its detection will be associated with the same event as the face.

  If there is no face detected in the image, no human body detections will be returned. Thus there will be no human body detection returned if detect_face is disabled.

Whether to detect faces on the image.

Whether to detect human bodies on the image.

Whether to estimate head pose in the image. The head pose is represented by pitch, yaw and roll angles.

Not supported with warped images (see warped_image or image_type parameter).

Whether to estimate emotions in the image. The probability of the following emotions is estimated: "anger", "disgust", "fear", "happiness", "sadness", "surprise", "neutral".

Whether to estimate mask in the image. The probability of the following masks states is estimated: "medical_mask", "missing", "occluded".

The estimated probability with the highest score is returned in the "predominant mask" parameter.

Whether to estimate glasses in the image. One of the following parameters is returned: "eyeglasses", "sunglasses", "no_glasses".

Whether to estimate liveness in the image. The "quality" and "score" probabilities are estimated. Based on these probabilities, the following result is returned: "real", "spoof", "unknown".

Liveness is utilized for this estimation. See "Administrator’s manual" for details about Liveness requirements.

⚠ Webcam or selfie photos are required. Otherwise the result may be incorrect.

⚠ Liveness licensing feature is required to be enabled to perform liveness estimation.

Liveness estimation is not supported for samples (warped images). The image_type parameter should be set to "0".

Whether to estimate deepfake feature on the image.

⚠ Deepfake licensing feature is required to be enabled to perform deepfake estimation.

Not supported with warped images (see warped_image parameter).

Whether to detect basic 68-point facial landmarks in the image.

Not supported with warped images (see warped_image or image_type parameter).

Whether to detect basic 5-point facial landmarks in the image.

Not supported with warped images (see warped_image or image_type parameter).

Whether to estimate input face image quality. Each of the following probabilities is estimated: "darkness", "light", "blurriness", "illumination", "specularity".

The quality factor is estimated for each probability. It's value is in the range [0..1] where 0 corresponds to low quality and 1 to high quality.

Note. Face recognition is not greatly affected by uneven illumination or high specularity.

Whether to estimate gaze in the image. The gaze is represented by yaw angle and pitch angle for both eyes at once.

Not supported with warped images (see warped_image or image_type parameter).

Whether to estimate eye attributes in the image. The "right_eye" and "left_eye" parameters are estimated, which contain the "state" (opened, closed or occluded) and "iris_landmarks" parameters.

Whether to estimate mouth attributes in the image. Each of the following probabilities is estimated: "opened", "occluded", "smile", "score".

Whether to estimate face descriptor(s).

Whether to estimate face basic attributes (gender, age, ethnicity).

Whether to create human body descriptor(s).

Whether to estimate human upper body on the image (headwear, clothing color, sleeve length).

Whether to estimate human lower body on the image (lower garment, shoes).

Whether to estimate human body accessories on the image (backpack presence).

Whether to estimate human body basic attributes on the image (age, gender).

Whether to create a human body warped image in the Base64 format.

Whether to create the face warped image in the Base64 format.

Whether to aggregate the following parameters:

• face descriptor
• body descriptor
• face basic attributes (age, gender)
• body basic attributes (apparent age, apparent gender)
• mask states
• emotions
• liveness
• deepfake
• upper body attributes
• lower body attributes
• accessories

Aggregation will be performed only if the appropriate parameters for estimating the listed above parameters are enabled.

If set, all estimated parameters from each image will be aggregated and displayed in the aggregate_estimations field of the response body as single objects for all images.

Otherwise, all estimated parameters from each image will be displayed in the estimations field of the response body as separate objects for corresponding images.

Maximum deviation pitch angle from 0. If the estimated value is equal or less than the threshold, the image is filtered.

Remove this parameter from the query, or set the parameter value to 180 if pitch angle filtration is not required.

When the value is set to "0", all the images are filtered.

Maximum deviation roll angle from 0. If the estimated value is equal or less than the threshold, the image is filtered.

Remove this parameter from the query, or set the parameter value to 180 if roll angle filtration is not required.

When the value is set to "0", all the images are filtered.

Maximum deviation yaw angle from 0. If the estimated value is equal or less than the threshold, the image is filtered.

Remove this parameter from the query, or set the parameter value to 180 if yaw angle filtration is not required.

When the value is set to "0", all the images are filtered.

Descriptor quality score threshold. The higher the quality score for the image, the better the extraction results.

All the attributes with a quality score below the threshold will be ignored (and not stored in the DB). The function will proceed as usual with all the remaining descriptors (if left).

Consult VisionLabs about the recommended value of this parameter.

Note! This parameter is not related to the image quality estimated in the detect request (estimate_quality request parameter).

Filter by mask states.

Filter by liveness states.

Filter by deepfake states.

Whether to try to auto orient image based on EXIF data.

The processed image should have EXIF data, otherwise auto orientation will not be performed.

Has no effect with tiff images (they're always auto oriented).

Ignored with warped images (see warped_image or image_type parameter).

Accept request header.

External request ID. Helps to uniquely identify messages in system logs that correspond to particular requests. If it was not set, the system will set it to the default format ("timestamp, UUID"). It will be returned with the response.