Facial Validation

In this section you will find the explanation of the basic principles of facial validation video analytics. Unlike classic facial recognition analytics for access control, facial validation analytics is much more advanced in its operation.
Principle of Operation
This video analytics tool is designed to generate specific alerts when individuals not registered in a facial recognition database cross a pre-designated area. Unlike traditional facial recognition systems, which trigger events by identifying people already registered in the database, this system operates in reverse. In other words, alerts are triggered when it detects individuals who are not part of the database,
Opening Configuration Parameters
To open the analytics configuration menu, stop all analytics that are currently running, and 1) click on the analytics configuration icon 2) select the analytics configuration file 3) click to expand the analytics corresponding to the desired camera.

Locate the parameter named 'face_validation_config_params' and click to expand it,

Setting Up Facial Validation Zone
The parameter 'face_validation_zone' determines the area where faces should be located for the system to analyze them. Click the 'Select Zone' button to open a view of the camera video and configure the zone,

The video view should open on the right side and from there you can create the facial validation area. Remember that this area is where the faces that you want the system to validate must pass.
If the video does not open, please check the camera settings, as there may be an issue connecting with the camera.
Setting Up Facial Validation Zone
The parameter 'face_validation_zone' determines the area where faces should be located for the system to analyze them. Click the 'Select Zone' button to open a view of the camera video and configure the zone,

The video view should open on the right side and from there you can create the facial validation area. Remember that this area is where the faces that you want the system to validate must pass.
If the video does not open, please check the camera settings, as there may be an issue connecting with the camera.
Explanation Configuration Parameters
PENDING: Al hacer click en el parámetro de face_validation_config_params se muestran los parámetros de configuración de la analítica,

A continuación se explican los más relevantes y los que en algún momento el usuario puede desear modificar para sintonizar el comportamiento del sistema.
-
enable: set to 1 to activate the analytics (always leave it at 1)
-
show_positive_recognition: set to 1 if you want the analytics to show positive recognition events when someone is recognized in the database. If set to 0, the analytics will not show positive facial recognition events.
-
show_negative_recognition: set to 1 if you want the analytics to show negative facial recognition events, that is, when a person was not recognized in the database. It is recommended to always leave it at 1.
-
user_lost_for_negative: set to 1 if you want the analytics to validate people who are not in the database when they have disappeared from the camera view. Set to 0 if you want the analytics to validate people who are not in the database when they are still in camera view.
-
bbox_min_size: Minimum length in pixels that a face must have to be considered in the validation. It is recommended to leave it at 30.
-
min_frames_in_zone_for_negative:number of samples that the face must be present in the video to be validated as a negative face. This parameter must be calibrated depending on the camera perspective. You can start with a value of XXXXX and observe the behavior. If you see the faces turn red too quickly increase it until you see the face turn from yellow to red in enough time. The idea is that this value is as large as possible without the faces leaving the camera view without being validated in red or green.
-
num_frames_lost_for_negative: number of samples so that a face is out of view of the video and generates an alert if it is not validated. It is recommended to leave this value at 20.
-
similarity_match_percentage: Minimum similarity percentage that a face from the video must have with a face from the database to be considered a facial recognition event. It is recommended to leave it at 50 and calibrate according to the place where the analysis is installed. The smaller the value, the more likely the system will make positive recognitions, but it is also possible that it will make associations with wrong people or false positives. The larger the value, the more strict the system will be to recognize people in the database, but it is also possible that people who are in the database will not be recognized.
Export Database to File
If you want to export an existing database to an Excel file, you can do so from the Database Editor. Select a name and click Export File

This will create a folder in Documents with the database data.
Run Video Analytics
After making all changes to the parameters or database, click the lower green Save button.

Click on the 'Play' button below. If the system is configured correctly, you should see the camera view with the facial validation area.

At this moment, we recommend conducting the following tests to verify the correct functioning of the system.
Have a person with a record in the database walk towards the validation area. The system should recognize them and change the face box color from yellow to green. If it doesn't recognize them, please verify the following:
The image in the database is frontal and meets the recommended specifications.
The similarity match percentage is not too high. You can verify this if the person stands in front of and close to the camera. If it recognizes them in that case, the percentage might be high. However, we do not recommend lowering it below 50% because it increases the possibility of false positives.
Have a person without a record in the database walk towards the validation area. After some time, the system should change the face box color from yellow to red.
If you encounter any issues, please contact our technical support team