1. Demo instruction:
The main interface of SDK Demo is shown in Figure 1-1. The main functions includes 1:1, 1:N, face database management, and setting options. The specific instructions are as follows
Pic 1-1 Demo main interface
(1) 1:1 function: it provides the similarity scores for two pictures, which is obtained by comparing with pic1 and pic2. As shown in Figure 1-2, click on the blank image to enter the image selection interface first and select the face image to be compared. Note: The face picture is in positive direction, and the recommended size is below 1920 X 1080, and the face is clear.
Pic 1-2 1:1 interface
(2) 1:N function: Please do the face registration in face database management first before using this function. Click “1:N” button as shown in Figure 1-3 after the registration is completed. The figure shows the result of comparing the showed face with the faces in the database. The upside score indicates the comparison threshold, the red box indicates the non-liveness, green box indicates the liveness face, and the upper left corner of the face frame indicates the registered name.
Pic 1-3 1:N main interface
(3) Face database management function: the main interface is shown in Figure 1-4. The main function includes face registration, face deletion, registered face display, face search, face page display. Click “Photo taking” to enter the interface as shown in Figure 1-5, you can fill in the registered name first in the upside box of interface, and then click the Camera pic below, which is the photo registration, as shown in Figure 1-6. After registration, please return to the face display interface and click the “registration” button to register. You can do 1:N again to do face comparison, the result is shown in Figure 1-3, which shows similarity scores, face names, and liveness or not, and so on.
Pic 1-4 the face frame management
Pic 1-5 face registration 1
Pic 1-6 face registration 2
(4) Settings: The interface is shown in Figure 1-7. The batch import function is described as follows: Click the “Please select import folder” button, you can enter the interface as shown in Figure 1-8, and select the corresponding image folder, and click “Batch Import”, which can start the import of the faces. After that, you can enter “the face frame management” to view the imported faces information.
Pic 1-7 Setting
Pic 1-8 picture selection
2. Algorithm Process
The Deepcam SDK uses the “deepcam.key” network authentication method. Therefore, you need to obtain the key file provided by Deepcam. For more details, please refer to the official website registration process at https://www.deepcam.cn/ or contact the staff directly.
The Deepcam algorithm interface only supports synchronous calls. The algorithm description flow chart is shown in Figure 2-1. The description is as follows:
(1) Entering the application, and loading all the feature objects which saved locally into the memory.
(2) Entering 1:N after registering the face information, and then turn on the camera to turn on the camera preview.
(3) Opening two threads, the thread 1 is responsible for face detection and camera picture echo, the thread 2 is responsible for liveness detection, feature extraction and 1:N comparison.
(4) Thread 1: it would do the face detection first, and then it will save the face information to the array of global variables, and update the camera picture preview at the same time.
(5) Thread 2: The liveness detection would be done first if you turned on this function, and if not, it would do feature extraction directly. 1:N matching will be done according to the extracted feature values, which can acquire many feature objects information such as the name and score.
(6) Finally, the required information will be displayed to the main interface.
Pic 1-2 Algorithm Process