Are you using FaceStation F2 v1.x.x and using the visual face as a credential? Please check this article if you want to upgrade to the firmware version to v2.x.x.
Overview
To improve the usability of customers, the criteria of the algorithm for extracting Visual Face templates from photos have been updated while improving the Visual Face performance. Accordingly, the face template stored in the FSF2 v1.x.x is not compatible with the FSF2 v2.x.x.
| Algorithm | Equipped On | Detect Face | Generate Warped Image | Extract Template |
| Old Algorithm | FaceStation F2 v1.x.x | Y | Y (250 x 250 px) | Y |
| New Algorithm (for GPU) | FaceStation F2 v2.x.x | Y | Y (112 x 112 px) | |
| Y |
To understand Visual Face extraction improvements, please refer to the article below:
Upgrade Notice - FaceStation F2
If you upgrade the firmware version from v1.x.x to v2.x.x, upgrade to v2.1.1 or later.
You might experience problems during firmware upgrades if you use firmware lower than v2.1.1.1. Notice
- Upgrading FaceStation F2 to v2.x.x version does NOT automatically check the compatibility of Visual Face.
2. How to Manage Visual Face Migration
The simplest way is to re-register a user with a new device. If it is difficult to re-register users, please refer to the methods below.
Case 1
- If you are managing raw image data or photo in a database, extract the template from the raw image data after upgrading the firmware to v2.x.x.
- In this case, you should extract and store new face template data to the user. Please refer to the article below for how to extract the visual face template.
[Both SDK] How to extract the visual face template from the image?
Case 2
If you do not have raw image data or the user's photo, please check the solution below.
This solution is to get the image data from the 1.x.x version and then extract the face template from the 2.x.x version.
- Device using v1.x.x firmware
1.1. Enable Image log and activate it with the user authentication success.
1.2. Request to authenticate the user to store the image log data accordingly.
1.3. Get the image log data with the user IDs.
1.4. Manage the Image log data in your application. - Device using v2.x.x firmware
2.1. Create a warped image from the image data using the Normalize function. This function needs to operate in the v2.x.x device.
(BioStar 2 Device SDK: BS2_GetNormalizedImageFaceEx / Suprema G-SDK: Normalize)
2.2. Extract the template from the warped image.
(BioStar 2 Device SDK: BS2_ExtractTemplateFaceEx / Suprema G-SDK: Extract)
2.3. Update the warped image data and face template data to your system in accordance with the user IDs.