ml.py

import numpy as np
import keras, os
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import *
from keras.optimizers import RMSprop

#np.set_printoptions(threshold=np.nan)

if __name__ == "__main__":

    # load in training data

    train_data = []
    train_files = os.listdir("train/arr")
    for file_name in train_files:
        train_data.append(np.loadtxt("train/arr/" + file_name))

    train_data = np.array(train_data)
    train_expected_outputs = np.loadtxt('train/outputs.txt')

    x_train = train_data.reshape(len(train_files), 32, 32, 1)
    x_train = x_train.astype('float32')
    print(str(x_train.shape))
    #print(str(x_train))

    y_train = keras.utils.to_categorical(train_expected_outputs, 2)
    print(str(y_train.shape))
    #print(str(y_train))

    # load in testing data

    test_data = []
    test_files = os.listdir("test/arr")
    for file_name in test_files:
        test_data.append(np.loadtxt("test/arr/" + file_name))

    test_data = np.array(test_data)
    test_expected_outputs = np.loadtxt('test/outputs.txt')

    x_test = test_data.reshape(len(test_files), 32, 32, 1)
    y_test = keras.utils.to_categorical(test_expected_outputs, 2)

    model = Sequential()
    model.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=(32, 32, 1)))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Flatten())
    model.add(Dense(128, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(2, activation='softmax'))
    model.compile(optimizer='sgd', loss='categorical_crossentropy', metrics=['accuracy'])
    model.fit(x_train, y_train, batch_size=2, epochs=25, verbose=1)

    score = model.evaluate(x_test, y_test, verbose=1)
    print('Test loss:', score[0])
    print('Test accuracy:', score[1])

    json_string = model.to_json()
    open('model.json', 'w').write(json_string)
    model.save_weights('weights.h5')

    for x in range(len(test_files)):
        prediction = model.predict(x_test[x].reshape(1, 32, 32, 1))
        print(str(np.argmax(prediction)))