-
To equitment and run the code you could view the original project at " https://github.com/jaisanant0/face-recognition-cnn "
-
My ideas is training model for 1-10 images/classs that i can appreciate the result thought My original dataset.
-
Then I had got a statistics result based on by visual such as the face-recognize was good activation with 4 images/class with by 85% accurately.
-
Conclusively, I recommend with meta learning when you use a pre-training you need only a little proper database, when more it not good for the algotithms.
-
The exactly of the method was arange 75-90% if you choose a good model as well as the good database and the good distancely calculation that the author use Euclip distance.
-
If you need a algorithms have more accurate, I think meta learninig needed to innovate that is my opinion
hoangkhoi13/Face-recognize-with-Meta-learning
In this project we reaching an approach as META learning .This method had been used in much time.
Jupyter Notebook