def main():
with tf.Graph().as_default():
cnn = CNN(image_size=FLAGS.image_size, class_count=len(Channel))
images, labels = load_data(
'data/test/data.csv',
batch_size=FLAGS.batch_size,
image_size=FLAGS.image_size,
class_count=len(Channel),
shuffle=False)
keep_prob = tf.placeholder(tf.float32)
logits = cnn.inference(images, keep_prob)
accuracy = cnn.accuracy(logits, labels)
saver = tf.train.Saver()
init_op = tf.global_variables_initializer()
with tf.Session() as sess:
sess.run(init_op)
saver.restore(sess, os.path.join(LOG_DIR, 'model.ckpt'))
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
accuracy_value = sess.run(accuracy, feed_dict={keep_prob: 0.5})
print(f'test accuracy: {accuracy_value}')
coord.request_stop()
coord.join(threads)
train_labels)
test_images, test_labels = fetch_images_and_labels(TEST_DIR)
test_images, test_labels = shaffle_images_and_labels(test_images, test_labels)
cnn = CNN(image_size=FLAGS.image_size, class_count=len(CLASSES))
with tf.Graph().as_default():
x = tf.placeholder(tf.float32, [None, PIXEL_COUNT])
labels = tf.placeholder(tf.float32, [None, len(CLASSES)])
keep_prob = tf.placeholder(tf.float32)
y = cnn.inference(x, keep_prob)
v = cnn.cross_entropy(y, labels)
train_step = cnn.train_step(v, FLAGS.learning_rate)
accuracy = cnn.accuracy(y, labels)
saver = tf.train.Saver()
init = tf.global_variables_initializer()
with tf.Session() as sess:
sess.run(init)
summary_op = tf.summary.merge_all()
summary_writer = tf.summary.FileWriter(LOG_DIR, sess.graph)
for i in range(FLAGS.step_count):
for j in range(int(len(train_images) / FLAGS.batch_size)):
batch = FLAGS.batch_size * j
sess.run(train_step,
feed_dict={
