main.py

import numpy as np
from deepscribe import DeepScribe
import mnist
import feature_extraction
from matplotlib import pyplot
import neural_network
import analysis


def run():
    # Code to build resnet18:
    # X, y, classes = neural_network.get_x_y_arrays('imgs_gray.npy', 'labels_gray.npy', n='all')
    # X = np.load("output/imgs_gray.npy")
    # y = np.load("output/labels_gray.npy")
    # neural_network.train(X, y, "output/resnet18_073120.h5")
    analysis.show_reports("output/split_data_resnet18_080720_tf2_dataAug.npz", "output/label_encoding_073120.npy", "output/resnet18_080720_tf2_dataAug.h5")
    # analysis.get_training_curves("records/training_output_resnet18_080520_tf2_lastDropOut075.txt", 20, 213)

    # Code to build mnist model:
    # X, y = mnist.dict_to_np_arrays("output/img_data1.npy", "output/label_data1.npy") # run with -l max for all symbols

    # X = np.load("output/img_data_gray.npy")
    # y = np.load("output/label_data_gray.npy")
    # mnist.run_mnist(X, y, "output/MNIST_model_on_OCHRE_100_epochs")
    # mnist.show_classification_report(X, y, "output/MNIST_model_on_OCHRE_100_epochs")
    # mnist.get_training_curve("records/training_output_100_epochs.txt")

    # Code to test feature extraction:
    # feature_extraction.extract_features("output/vgg16_fts_color.npy")

    # Code to show feature maps - does not work:
    # X = np.load("output/vgg16_fts_color.npy")
    # print(X.shape)
    # for fmap in X:
    #     ix = 1
    #     for i in range(8):
    #         for i in range(8):
    #             ax = pyplot.subplot(8,8,ix)
    #             ax.set_xticks([])
    #             ax.set_yticks([])
    #             pyplot.imshow(fmap[0,:,:])
    #             ix += 1
    # pyplot.show()

    # Code for random forest:
    # y = np.load("output/label_data_gray.npy")
    # feature_extraction.random_forest(X, y)


if __name__ == '__main__':
    run()