preview_predictions.py

import readsample as rs
import numpy as np
import display


def preview_reconstructed(name):
    x_test_input_unchanged = rs.read_images_from_pkl('validation_input.pkl')
    imgs = rs.read_images_from_pkl(name)

    n = 5
    np_images = []
    for i in range(0, n):
        img_array = imgs[i]
        center = (int(np.floor(img_array.shape[0] / 2.)), int(np.floor(img_array.shape[1] / 2.)))
        target = img_array[center[0] - 16:center[0] + 16, center[1] - 16:center[1] + 16, :]
        np_images.append(target)

    display.show_reconstructed_images(x_test_input_unchanged, np_images, True)


def preview_middle(name):
    x_test_input_unchanged = rs.read_images_from_pkl('validation_input.pkl')
    reshaped_decoded_imgs = rs.read_images_from_pkl(name)

    display.show_reconstructed_images(x_test_input_unchanged, reshaped_decoded_imgs, True)


def preview_predictions(name):
    imgs = rs.read_images_from_pkl(name)
    display.show_images(imgs, 5, True)


def main():
    # preview_middle('autoencoder_decoded.pkl')
    # preview_predictions('gan_decoded.pkl')
    # preview_predictions('convolution_decoded.pkl')
    preview_predictions('./tmp/gen_pre_train_decoded.pkl')
    # preview_full('./tmp/gen_pre_train_decoded.pkl')


if __name__ == "__main__":
    main()