146 |
P a g e
User.1.2 is the second photo taken of the 100 total photos of the user with the id number 1. Figure 5 shows part of the program for
creating the datasets for the enhanced training.
Figure 5. Part of the program that shows the algorithm to create datasets with face ids.
Initiating the Enhanced Training
After
the datasets have been created, we now have 100 grayscale pictures of
someone’s face which we will use for the enhanced
training. There are 3 unique faces used in the Enhanced training, so each of the 3 faces will have 100 grayscale images. Figure 6
shows the dataset created from running the program in Figure 5.
Figure 6. The dataset created from the program shown in Figure 5.
After the datasets have been made, we will then put the datasets to the training algorithm to initiate the enhanced training. Firstly,
we must declare
in the training program, the path to reach the datasets we made by filling the name
of the folder in which the
datasets are stored in the variable path. Secondly, we will then create the variable recognizer
and fill the function
cv2.face.LBPHFaceRecognizer_create() so that we can
store and train the samples after we retrieve the datasets and its’ ids.
Thirdly, we will call the pre-trained classifier again to help the training algorithm to recognize the faces in the datasets as seen on
Figure 7. Finally, we will call the getImagesAndLabels class to get the face samples with their ids accordingly in which we will finally
be able to train and finally save the enhanced training in the trainer file with name and format of trainer.yml as seen on Figure 8.
Figure 7. Part of the training program that shows the path as well as declaring the recognizer and detector variable.
