Page History

x32: (x32)\FRS
x64: (x64)\FRS

The key defines on how many CPU cores the face recognition module will operate (firserver.run process). All cores are in use by default.

The key sets the timeout upon which Face Server ‘loses’ the face if there is no new data about it.

Sets the parameters of generating and attaching face biometrics in the detection tool:

1 – Face detection tools generate and attach face biometrics to the photo. 

0 – Face detection tools do not generate face biometrics - the face is only captured.

Note. Face recognition Server that performs search in the user database checks up whether it has attached biometrics when it gets a face photo. If yes, then biometrics are not further generated for search in the database - data attached to the photo is in use. If no, then recognition Server generates biometrics itself using the photo. 

Specifies what data will be stored in the database when using the Face recognition Server:

1 - all captured faces.
0 - only the faces about which there are recognition results.

The key is in use in order to avoid the database overflow.

Note. Since version 6.2 use the UseFaceDB key in order to minimize the amount of data stored in DB.

The key is used to configure the face detection tool – it sets time on the expiry of which the face is considered lost.

For instance, if timeUntilLost = 1 and a man covers his face with the hands in front of the camera and then shows it again, then the face detection tool does not capture it as it considers that the face was lost for the time it was covered.

The key sets the mode for storing metadata, images and vectors of captured and reference faces.

0 – metadata, images and vectors are stored only in database.

1 – metadata, images and vectors are stored in database, images and vectors are also storing in a folder on disk. Path to folder is specified using the FaceDB parameter(it is possible to specify both a local and network folder).

2 – metadata are stored in the database, images and vectors are storing only in a folder on disk. Path to folder is specified using the FaceDB parameter (it is possible to specify both a local and network folder).

CpuCoreCountAddPerson

The key defines the number of CPU cores where the vectors of persons are generated when adding a reference face to the database.

0 - vectors are generated on the number of cores specified in the CpuCoreCount key. The cores are used successively. Recognition sample is bound to each core. 

1 - only 1 fixed core (different from those specified in CpuCoreCount) is in use when adding a reference face to the database. 

2 - only 2 fixed cores (different from those specified in CpuCoreCount) are in use when adding a reference face to the database.

Face.HardwareCPU

The key defines the order of binding streams to CPU cores when generating vectors of persons.

no value or 0 - streams are bound randomly to CPU cores.
1 - streams are bound successively to CPU cores depending on their total number. 
2 - streams are bound to first cores of each CPU.

Mechanism of displaying faces. When a new mechanism of displaying is enabled, the photo in the list of captured faces is not duplicated - it is updated in the same cell while the tracker "sees" it.  

0 - new mechanism of displaying faces. 

1 - old mechanism of displaying faces. 

The key enables saving the last recognized face and all its necessary data to a file. This image file can be sent directly to the recognition server via the iidk_client utility.

0 - General operation mode.

1 - When the face is captured, the captureFrame (cam_N) .log file is created (where N is the camera number that captured the face) in the <Intellect Installer directory>\Modules (64)\ folder. This file will contain the last recognized face and all its necessary data. This image file can be sent directly to the recognition server via the iidk_client utility.

Note. The file is overwritten every time a new face is captured.

The key restricts resolution (width) of the photo when adding it to the faces database.

Set greater value to the key if the following message appears when adding a face to DB: "Adding error. The face was not added. Insufficient image quality to add".

The key restricts resolution (height) of the photo when adding it to the faces database.

The key enables selecting the Tevian algorithm.

0 – less resource intensive algorithm is in use

1 – more resource intensive algorithm is in use 

Important! Biometric vectors received using this algorithm cannot be used with alternative one. For proper operation of Face Intellect after changing this parameter biometric vectors are to be converted as described in Transition between face recognition modules section. 

An alternative algorithm for filtering false triggerings of the tracker. This key is in use if Tevian.GlobalTrackingDetectionFilter=0 and Tevian.DetectionFilterValue > 0 (see the corresponding parameters below). 

0 - ALG1

1 - ALG2

Note. It is not recommended to change this parameter without consulting Tevian.

Configuring alternative algorithm filtering. It is in use if the value of Tevian.GlobalTrackingDetectionFilter is more than 0 (see the corresponding parameter below).

Note. It is not recommended to change this parameter without consulting Tevian.

Activates an additional check to filter out false triggerings of the tracker. If faces in masks are not recognized, it is recommended to disable this check, however, this can cause false triggerings of the tracker.

0 - disabled

1 - enabled.

Note. It is not recommended to change this parameter without first consulting with Tevian.

Face detection operation mode (identification). The recommended parameter value is 0.

0 - ALG1

1 - ALG2

Note. It is not recommended to change this parameter without consulting Tevian.

This parameter affects the detector captioning capacity. If there is a great value, the tracks can rend.

Note. It is not recommended to change this parameter without consulting Tevian.

Face detector confidence threshold (do not confuse with face quality).

Note. It is not recommended to change this parameter without consulting Tevian.

The key enables selecting the Tevian algorithm for recognition and face search.

0 - ALG1 less resource-intensive algorithm is in use.

1 - ALG2 identification of the higher quality, but more resource-intensive.

2 - PRECISE the highest quality of recognition, but the most resource-intensive. This algorithm is recommended if significantly less time is spent on face recognition than on detection (if there are few people in the frame).

Important! Biometric vectors received using this algorithm cannot be used with any other one. For proper operation of Face Intellect after changing this parameter biometric vectors are to be converted as described in Transition between face recognition modules.

The Specifies the number of seconds to accumulate the best frames of a track for recognition in the track. The longer the time, the poorer is the accuracy of the age determininationincrease in time makes it possible to average out the recognized face characteristics, but it can result in the poorer accuracy of the face characteristics recognition.

Note. For example, if the 1 second value is set, the average error in determining the age will be 2-3 years, and if the 5 seconds value is set, it will be 3-4 years.

Time since the last recognition after which the track is considered ended.

Note. It is not recommended to change this parameter without consulting Tevian.

Face.PollTime

The key specifies the polling time of the HUAWEI server in milliseconds.

Note. Face-Intellect requests the events only since the time it is launched. If events contain an earlier time than the launch time of Face-Intellect, these events will be ignored. That is why the HUAWEI server time and the Face-Intellect time should be synchronized.