eInfochips iMX8XML, Руководство пользователя

Страница: 27 / 53

detected faces - 0
[]
detected faces - 0
`q` pressed , Exiting... <--- User Press “q” here from keyboard attached to board
Total Elapsed time: 31.82 sec
Directory anil removed
Quantifying faces from training dataset...
Processing Image : training_data/unknown/00000022.JPG
Processing Image : training_data/unknown/00000034.JPG
Processing Image : training_data/unknown/ellie_sattler.jpg
Processing Image : training_data/unknown/00000009.jpg
Processing Image : training_data/unknown/00000004.jpg
Encoding done for image = training_data/unknown/00000022.JPG
Encoding done for image = training_data/unknown/00000034.JPG
Processing Image : training_data/unknown/ds3.jpg
Processing Image : training_data/unknown/00000016.jpg
This will re-train model with new label “anil” training images along with other training images which
are already provided inside training_data folder.
Here we create new training directory anil or copy new images inside existing directory if already have
directory with same name. Therefore, if we have low confidence for any face we can re-train our
model with that face and can increase confidence for that.
Training takes few minutes (around 10-15 minutes) depending on number of training images. In
training, we utilize four cores of CPU and do training on images using four concurrent python process,
which speed up our executions.
If User simply want to re-train model without providing any new input images or labels then we can
do that by simply press “enter” and leave blank when script ask for new label . This scenario is only
useful if your existing train model is corrupted or deleted by mistake. We can do training on our HOST
Linux machine and can use trained model for testing but python module version must be same or
compatible with version
of board’s package.