Autoencodeur module

Autoencodeur.import_images(nb_images, start, sexe, filesnames)[source]

To give numerotation to the images, import images of the dataset and import nb_images, strat with the images number start and take the same number of male and female.

input :
nb_images (int) :

Number of images we want Retrain start at the image start to import

start (int) :

Import start at the image start

sexe (list) :

List of the sex -1 (Female) or 1 (Male)

filesnames (list) :

List of all the filesnames we want to import

output :
x_data (numpy array) :

List of the images in their np.array form

Autoencodeur.load_attr()[source]

Create the liste of filesnames and sexe of the images

output :
sexe (list) :

List of the sex -1 (Female) or 1 (Male)

filesnames (list) :

List of all the filesnames we want to import

Autoencodeur.retrain(sexe, filenames, nb_images, start, plot=0)[source]

Fit the autoencodeur with nb_images new images, start with the start images of the dataset. Finally save the autoencodeur with the number num.

input :
sexe (list) :

List of the sex -1 (Female) or 1 (Male)

filenames (list) :

List of all the filesnames we want to retrain

nb_images (int) :

Number of images we want to retrain

start (int) :

Retrain start at the image start

plot (int) :

If 1 plot the result of the autoencoder (compare before and after autoencoder)