main.py

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import image

# LOAD DATA
mnist = input_data.read_data_sets('data/', one_hot=True)
images = image.run()

# INIT WEIGHTS
def init_weights(shape):
	init_random_dist = tf.truncated_normal(shape, stddev=0.1)
	return tf.Variable(init_random_dist)

# INIT BIAS
def init_bias(shape):
	init_bias_vals = tf.constant(0.1, shape=shape)
	return tf.Variable(init_bias_vals)

# CONV2D
def conv2d(x, W):
	return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')

# POOLING
def max_pool_2by2(x):
	return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')

# CONVOLUTIONAL LAYER
def convolutional_layer(input_x, shape):
	W = init_weights(shape)
	b = init_bias([shape[3]])
	return tf.nn.relu(conv2d(input_x, W) + b)

# NORMAL (FULLY CONNECTED)
def normal_full_layer(input_layer, size):
	input_size = int(input_layer.get_shape()[1])
	W = init_weights([input_size, size])
	b = init_bias([size])
	return (tf.matmul(input_layer, W) + b)

# PLACEHOLDERS
x = tf.placeholder(tf.float32, shape=[None, 784])
y_true = tf.placeholder(tf.float32, shape=[None, 10])

# LAYERS
x_image = tf.reshape(x, [-1, 28, 28, 1])
convo_1 = convolutional_layer(x_image, shape=[6, 6, 1, 32])
convo_1_pooling = max_pool_2by2(convo_1)
convo_2 = convolutional_layer(convo_1_pooling, shape=[6, 6, 32, 64])
convo_2_pooling = max_pool_2by2(convo_2)
convo_2_flat = tf.reshape(convo_2_pooling, [-1, 7*7*64])
full_layer_one = tf.nn.relu(normal_full_layer(convo_2_flat, 1024))

# DROPOUT
hold_prob = tf.placeholder(tf.float32)
full_one_dropout = tf.nn.dropout(full_layer_one, keep_prob=hold_prob)
y_pred = normal_full_layer(full_one_dropout, 10)

# LOSS FUNCTION
cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_true, logits=y_pred))

# OPTIMIZER
optimizer = tf.train.AdamOptimizer(learning_rate=0.0001)
train = optimizer.minimize(cross_entropy)

# INITIALIZE VARIABLES
init = tf.global_variables_initializer()
steps = 5000
saver = tf.train.Saver()

# RUN SESSION
def run():
	with tf.Session() as sess:
		# SESS_RESTORED = False


		try:
			saver.restore(sess, './models/mnist_model.ckpt')
			SESS_RESTORED = True

		except:
			print('\n\n\n\t\tERROR UNABLE TO RESTORE SESSION\n\n\t\tNOW EXITING PROGRAM\n\n\n')
			SESS_RESTORED = False
			quit()

		if SESS_RESTORED is False:
			sess.run(init)
			for i in range(steps):
				batch_x, batch_y = mnist.train.next_batch(50)
				# batch_x, batch_y = next_batch(50)
				sess.run(train, feed_dict={x: batch_x, y_true: batch_y, hold_prob: 0.5})

				if i % 100 == 0:
					print('ON STEP: {}'.format(i))
					print('ACCURACY: ')
					matches = tf.equal(tf.argmax(y_pred, 1), tf.argmax(y_true, 1))
					acc = tf.reduce_mean(tf.cast(matches, tf.float32))
					print(sess.run(acc, feed_dict={x: mnist.test.images, y_true: mnist.test.labels, hold_prob: 1.0}))
					print('\n')

					for i in range(5):
						print(f'PRED: {tf.argmax(y_pred, 1).eval(feed_dict={x: mnist.test.images, y_true: mnist.test.labels, hold_prob: 1.0})[i]} TRUE: {tf.argmax(y_true, 1).eval(feed_dict={x: mnist.test.images, y_true: mnist.test.labels, hold_prob: 1.0})[i]}')

					print('\n')

				saver.save(sess, './models/mnist_model.ckpt')

		else:

			row = []
			prediction = tf.argmax(y_pred, 1)
			
			for i in range(len(images)):
				pred = sess.run(prediction, feed_dict={x:images[i], hold_prob: 1.0})
				row.append(str(pred[0]))

				# print(f'IMAGE: {i}     PRED: {pred[0]}    TYPE: {type(pred[0])}')

				if len(row) == 3:
					val = int(''.join(row))
					
					print(f'ROW: {int((i+1)/3)}    VALUE: {val}')

					if val % 3 == 0:
						if val % 5 == 0:
							print('fizzbuzz')

						else:
							print('fizz')

					elif val % 5 == 0:
						print('buzz')
						
					print('\n')
					row = []


if __name__ == '__main__':
	run()