Overview


  • If you register a visual face as a credential or store the visual face template on the smart card, you need to extract a face template.

  • If you scan a face directly through the device, the template data is extracted during the scan.

  • If you use an image file, you need to extract template data through several processes.


Extracting visual face templates is only possible with FaceStation F2 and BioStation 3.

Instructions on how to get a face template from the image

To extract a face template using the SDK, please follow the below steps.
(Both the BioStar 2 Device SDK and the Suprema G-SDK are the same.)


Step 1: Prepare the image file

  • Please prepare an image file for registration. The photo format should be JPG and PNG.
  • How to choose a proper image for user face credential:

 

Step 2: Normalize the image using the device

  • Terminology "Normalization" is a kind of generalization that extracts the face from the physical image(raw image) which contains different body parts.
  • If you proceed with normalization, you can get a warped image.
  • There may be differences in warped image size for each firmware version.

    AlgorithmEquipped OnDetect FaceGenerate Warped ImageExtract Template
    Old AlgorithmFaceStation F2 v1.x.xYY
    (250 x 250 px)    		Y
    New Algorithm
    (for GPU)    		BioStar 2
    v2.9.0 and above    		Y
    * Same way    		Y
    (112 x 112 px)    		Y
    FaceStation F2 v2.x.x
    New Algorithm
    (for NPU)    		BioStation 3Y

  • Please refer to the article below for more information about image extraction.
    [BioStar 2] Improved Image Extraction for Visual Face Credentials on BioStar 2
  • Without a device, "Normalization" is impossible. This feature only runs on the device.
  • Create a warped image from the image file using the Normalize function.
    (BioStar 2 Device SDK:     BS2_GetNormalizedImageFaceEx / Suprema G-SDK: Normalize) 

What does "WARP" mean?
WARP is a kind of generalization that extracts the face part from the original image that may contain various body parts.


Step 3: Extract the face template from the warped image