def set_watched(self, id):
for item in [item for item in self._items if item['id']==id and item['playcount']==0]:
item['playcount'] = 1
item['lastplayed'] = helper.current_time()
self.update(['Suggested', 'Recent', 'Random'] if self.__randomUpdateMethod else ['Suggested', 'Recent'])
def start_playing(self, id):
for item in [item for item in self._items if item['id']==id]:
item['lastplayed'] = helper.current_time()
def train_nn(sess, epochs, batch_size, get_batches_fn, train_op, loss,
predictions, input_image, correct_label, keep_prob, learning_rate,
saver):
"""
Train neural network and print out the loss during training.
:param sess: TF Session
:param epochs: Number of epochs
:param batch_size: Batch size
:param get_batches_fn: Function to get batches of training data. Call using get_batches_fn(batch_size)
:param train_op: TF Operation to train the neural network
:param cross_entropy_loss: TF Tensor for the amount of loss
:param input_image: TF Placeholder for input images
:param correct_label: TF Placeholder for label images
:param keep_prob: TF Placeholder for dropout keep probability
:param learning_rate: TF Placeholder for learning rate
"""
# max_loss = 10000
for epoch_i in range(1, epochs + 1):
epoch_loss = 0
steps = 1
for images, labels in get_batches_fn(batch_size):
_, batch_loss = sess.run(
[train_op, loss],
feed_dict={
input_image: images,
correct_label: labels,
keep_prob: 0.75,
learning_rate: 0.0001
})
epoch_loss += batch_loss
mean_epoch_loss = epoch_loss * 1.0 / steps
# print every Epoch
print(
"Epoch {}/{} \t".format(epoch_i, epochs),
"Epoch Loss: {:.4f} Last Batch Loss: {:.4f}".format(
mean_epoch_loss, batch_loss))
steps += 1
# every 10 epochs, run on mini_test images, save test images with mask
if epoch_i % 10 == 0:
mean_iou = helper.save_inference_samples(runs_dir,
data_dir,
sess,
image_shape,
predictions,
keep_prob,
input_image,
epoch=epoch_i,
mini=True,
compute_iou=True)
print("Epoch{} iou:{} ".format(epoch_i, mean_iou))
# save last 3 model
if epoch_i > (epochs - 3):
save_path = saver.save(
sess,
"./models/models-{}-{}-{}/segmentation_model.ckpt".format(
epoch_i, mean_epoch_loss, current_time()))
# TODO:val_loss is decreasing, save model
# use last model test on whole testset
final_modeliou = helper.save_inference_samples(runs_dir,
data_dir,
sess,
image_shape,
predictions,
keep_prob,
input_image,
epoch=epoch_i,
mini=False,
compute_iou=True)
print("Final Model iou: ", final_mean_iou)