def testEagerMemory(self):
training_util.get_or_create_global_step()
logdir = self.get_temp_dir()
with summary_ops.create_file_writer(
logdir, max_queue=0,
name='t0').as_default(), summary_ops.always_record_summaries():
summary_ops.generic('tensor', 1, '')
summary_ops.scalar('scalar', 2.0)
summary_ops.histogram('histogram', [1.0])
summary_ops.image('image', [[[[1.0]]]])
summary_ops.audio('audio', [[1.0]], 1.0, 1)
def testEagerMemory(self):
training_util.get_or_create_global_step()
logdir = self.get_temp_dir()
with summary_ops.create_file_writer(
logdir, max_queue=0,
name='t0').as_default(), summary_ops.always_record_summaries():
summary_ops.generic('tensor', 1, '')
summary_ops.scalar('scalar', 2.0)
summary_ops.histogram('histogram', [1.0])
summary_ops.image('image', [[[[1.0]]]])
summary_ops.audio('audio', [[1.0]], 1.0, 1)
def testSummaryOps(self):
training_util.get_or_create_global_step()
logdir = tempfile.mkdtemp()
with summary_ops.create_file_writer(
logdir, max_queue=0,
name='t0').as_default(), summary_ops.always_record_summaries():
summary_ops.generic('tensor', 1, '')
summary_ops.scalar('scalar', 2.0)
summary_ops.histogram('histogram', [1.0])
summary_ops.image('image', [[[[1.0]]]])
summary_ops.audio('audio', [[1.0]], 1.0, 1)
# The working condition of the ops is tested in the C++ test so we just
# test here that we're calling them correctly.
self.assertTrue(gfile.Exists(logdir))
def testSummaryOps(self):
training_util.get_or_create_global_step()
logdir = tempfile.mkdtemp()
with summary_ops.create_file_writer(
logdir, max_queue=0,
name='t0').as_default(), summary_ops.always_record_summaries():
summary_ops.generic('tensor', 1, '')
summary_ops.scalar('scalar', 2.0)
summary_ops.histogram('histogram', [1.0])
summary_ops.image('image', [[[[1.0]]]])
summary_ops.audio('audio', [[1.0]], 1.0, 1)
# The working condition of the ops is tested in the C++ test so we just
# test here that we're calling them correctly.
self.assertTrue(gfile.Exists(logdir))
def testSummaryOps(self):
logdir = self.get_temp_dir()
writer = summary_ops.create_file_writer(logdir, max_queue=0)
with writer.as_default(), summary_ops.always_record_summaries():
summary_ops.generic('tensor', 1, step=1)
summary_ops.scalar('scalar', 2.0, step=1)
summary_ops.histogram('histogram', [1.0], step=1)
summary_ops.image('image', [[[[1.0]]]], step=1)
summary_ops.audio('audio', [[1.0]], 1.0, 1, step=1)
with self.cached_session() as sess:
sess.run(summary_ops.summary_writer_initializer_op())
sess.run(summary_ops.all_summary_ops())
# The working condition of the ops is tested in the C++ test so we just
# test here that we're calling them correctly.
self.assertTrue(gfile.Exists(logdir))
def testSummaryOps(self):
logdir = self.get_temp_dir()
writer = summary_ops.create_file_writer(logdir, max_queue=0)
with writer.as_default(), summary_ops.always_record_summaries():
summary_ops.generic('tensor', 1, step=1)
summary_ops.scalar('scalar', 2.0, step=1)
summary_ops.histogram('histogram', [1.0], step=1)
summary_ops.image('image', [[[[1.0]]]], step=1)
summary_ops.audio('audio', [[1.0]], 1.0, 1, step=1)
with self.cached_session() as sess:
sess.run(summary_ops.summary_writer_initializer_op())
sess.run(summary_ops.all_summary_ops())
# The working condition of the ops is tested in the C++ test so we just
# test here that we're calling them correctly.
self.assertTrue(gfile.Exists(logdir))
def main():
batch_size = 16
dataset = get_dataset(batch_size,
x_train,
y_train,
distorted_image_fn,
shuffle=True,
repeat=True)
model = ClassificationNet(st_out_dim=IMG_SHAPE, num_class=NUM_CLASS)
optimizer = tf.train.AdamOptimizer(learning_rate=0.001)
global_step = tf.train.get_or_create_global_step()
for i, (image, rotated_image, angle, label) in enumerate(dataset, start=1):
global_step.assign_add(1)
with summary.record_summaries_every_n_global_steps(10):
loss, prediction, transformed, theta = model.train(
optimizer, rotated_image, label)
acc = model.accuracy(prediction, label)
# test
if i % 500 == 0:
total_acc = 0
dataset_test = get_dataset(
1000, x_test, y_test,
distorted_image_test_fn).make_one_shot_iterator()
split = 10000 // 1000
for _ in range(split):
image_test, rotated_image_test, angle_test, label_test = dataset_test.get_next(
)
logits_test, transformed_test, theta_test = model(
rotated_image_test)
prediction_test = tf.nn.softmax(logits_test)
acc_test = model.accuracy(prediction_test,
label_test).numpy()
total_acc += acc_test
print(total_acc / split)
summary.scalar('accuracy/test', total_acc / split)
if i % 10 == 0:
print("step: {}, loss: {}, accuracy: {}".format(
int(global_step), float(loss), float(acc)))
summary.scalar('loss', loss)
summary.scalar('accuracy/training', acc)
# summary images
origin_images = image_utils.image_gallery(image.numpy(),
columns=4,
expand_dim=True)
rotated_images = image_utils.image_gallery(
rotated_image.numpy(), columns=4, expand_dim=True)
transformed_images = image_utils.image_gallery(
transformed.numpy(), columns=4, expand_dim=True)
summary.image('image/original', origin_images)
summary.image('image/rotated', rotated_images)
summary.image('image/transformed', transformed_images)
