tpot_classification module

class tpot_classification.TpotClassifier(data_path, gabor=False, reduction_method=None)

Bases: object

Trains a machine learning model with tpot, which can then be used to predict attractiveness of pictures of humans.

Parameters:

• data_path – Path to the folder that has the image folder and users.csv
• make_data_set – Takes the images from the data folder, finds faces in them and transforms those into 64x64 grayscale images. Making the data set takes a long time.
• gabor – This determines if gabor filter is used in training.
• reduction_method – Which method to use for dimensionality reduction. Supported types are ‘pca’ and ‘lda’.

build_filters()

returns a list of kernels in several orientations

static create_data(data_path)

Makes a 64x64 grayscale image dataset of faces.

Parameters:data_path – Path to the folder that has the image folder and users.csv

load()

Loads the fitted tpot models into usage

Returns:None

predict(image)

Predicts the attractiveness of a picture with a human in it.

Parameters:image – The image with the human in it. Returns:The prediction.

process(img, filters)

Returns the img filtered by the filter list.

Parameters:

• img – Image to be filtered.
• filters – The gabor filters to be used.

train()

Uses tpot to train a machine learning model to predict attractiveness of pictures of human faces.

Returns:Various accuracy measures.