The Facial Recognition Service enables users to capture face images for face registration, create face groups, adjust face threshold sensitivity, and activate additional face attributes such as gender and emotion for your Facial Recognition tasks.

Register new users

To register a user, go to the Setting page (1) and open Facial Recognition Service dialogue (2)

You will be prompted to the Facial Recognition Service dialogue. Press Add New User

Enter the User name, user ID

From the group name drop-down list, you can assign the user to a group or create a new group by entering the name directly into the dialogue.

You can also press Edit to create and manage groups

Press Save New User

Assigning a user to a group is optional. However, if the task requires identifying a specific individual, that person must be assigned to a group. A user can be assigned to multiple groups. These groups will then be displayed as objects for selection within the AI task

Now, you have added a new user to your list. Repeat the previous steps to add more users.

 

Assign face images for users

There are two ways to add face images for users.

Add face image from local storages

Select a user you want to assign face images to and click Add Images (1)

Select the face images from local drives and press OK. 

ANSVIS will automatically crop the image to face size if detectable and add them to the face database.

 

Add face image from face recording.

ANSVIS offers a face-recording feature that automatically detects and collects facial images from recorded videos, allowing you to quickly assign these images to any user on your list.

To do this, press Add Face Recording Tasks.

You will be prompted to the Face Recording Task dialogue

• Select the camera you want to record faces
• Recording time range: you can use this feature to schedule your recording task
• Maximum face images to record per ID: set how many images you want to capture per face ID
• Maximum number of face images to record: set how many images you want to capture for the whole recording
• Minimum face score: Set the sensitivity for face detection; the lower the face score, the more sensitive it is.
• Min face size: set the minimum face size to capture in pixel

Once you have configured all parameters, press Start Recording Faces. You will be redirected to the Face Recording Tasks Status dialogue. 

The criteria for stopping the task are either the recording time has elapsed, or the maximum number of face images has been reached. You can also stop the task manually by pressing the Stop button.

Once the recording task is finished, press View Images, and then the Facial Recognition Labelling dialogue will be opened.

• You can manually register each image (1) or select multiple images to register (2). If the user hasn't been created yet, you can create a new one by typing directly in the drop-down list. The registration will be saved when you click the Load Previous, Next, or Save buttons.
• You can also Export All Images (3) to local storage to organize and import them using the local image method
• After completing registration, press the Save And Exit button to close the dialogue.

You can always return to the task to register the remaining images. Pressing "Delete Task" will remove the recording and its images, but the registered images will not be affected.

New users and their facial images will be added to the face database. Henceforth, the face recognition system will be able to detect and identify these users when they enter the field of view.

Additional Settings

These settings will be applied to all face recognition tasks.

Min face size (px)

The Minimum Face Size allows users to define the smallest face dimensions (width and height) for detection. This ensures that the system only add and detects face larger than the specified size, reducing false positives for very small faces.

You can preview the overlay of the min face size by going to a face recording task setting

Known Person Threshold

The known person threshold acts like a security vs convenience slider. A higher threshold allows higher recognition sensitivity with fewer images but increases the risk of mistaken identity. Conversely, a lower threshold requires more images for enrollment but ensures more accurate recognition. This lets you tailor the system's balance between speed and security.

Face Attributes

You can enable gender, emotions, and head pose options to provide extra information on detected faces. 

To see this information in the event view, check the Extra Info box

Tips

You can also collect face images for registration from the event log. To do this:

• Go to the Event Log page of the Facial Recognition AI Task
• Right-click on the event list
• Select Export All Face Images

The Export Face Images dialogue will be opened:

• Choose a dedicated folder to export (1)
• Select the Unknown (Unregistered) Faces option (2)
• Click Export (3)

ANSVIS will auto-crop all exported images and return only the face area. Then, you can import these images for registration if needed.