def _prepare_augmented_batch(self, tensors, batch_size, image_size=None):
if self.subset == "train":
assert batch_size == 1, "only implemented for 1 so far"
augmentor_strs = self.config.unicode_list("augmentors_train", [])
augmentors = parse_augmentors(augmentor_strs, self.void_label())
for augmentor in augmentors:
tensors = augmentor.apply(tensors)
if image_size is None:
image_size = self.image_size
tensors = assemble_input_tensors(tensors, image_size)
# batchify
keys = tensors.keys()
tensors = {k: tf.expand_dims(tensors[k], axis=0) for k in keys}
else:
augmentor_strs = self.config.unicode_list("augmentors_val", [])
assert "scale" not in augmentor_strs, "scale augmentation during test time not implemented yet"
assert "translation" not in augmentor_strs, "translation augmentation during test time not implemented yet"
augmentors = parse_augmentors(augmentor_strs, self.void_label())
tensors = [tensors for _ in xrange(batch_size)]
tensors = [apply_augmentors(t, augmentors) for t in tensors]
tensors = [assemble_input_tensors(t) for t in tensors]
# batchify
keys = tensors[0].keys()
tensors = {
k: tf.stack([t[k] for t in tensors], axis=0)
for k in keys
}
return tensors
def _parse_augmentors_and_shuffle(self):
if self.subset == "train":
shuffle = True
augmentor_strs = self.config.unicode_list("augmentors_train", [])
augmentors = parse_augmentors(augmentor_strs, self.void_label())
if len(augmentors) == 0:
print >> log.v1, "warning, no data augmentors used on train"
else:
shuffle = False
augmentor_strs = self.config.unicode_list("augmentors_val", [])
augmentors = parse_augmentors(augmentor_strs, self.void_label())
return augmentors, shuffle
def jointly_augment_examples_after_resize(self, tensors_batch):
augmentors_str = self.config.string_list("augmentors_" + self.subset,
[])
augmentors = parse_augmentors(augmentors_str, self.config)
for aug in augmentors:
tensors_batch = aug.batch_apply_after_resize(tensors_batch)
return tensors_batch
def augment_example_after_resize(self, tensors):
augmentors_str = self.config.string_list("augmentors_" + self.subset,
[])
augmentors = parse_augmentors(augmentors_str, self.config)
for aug in augmentors:
tensors = aug.apply_after_resize(tensors)
return tensors
def _prepare_augmented_batch(self, tensors, batch_size, image_size=None):
if self.subset == "train":
assert batch_size == 1, "only implemented for 1 so far"
augmentor_strs = self.config.unicode_list("augmentors_train", [])
augmentors = parse_augmentors(augmentor_strs, self.void_label())
for augmentor in augmentors:
tensors = augmentor.apply(tensors)
if image_size is None:
image_size = self.image_size
tensors = assemble_input_tensors(tensors, image_size)
# batchify
keys = list(tensors.keys())
tensors = {k: tf.expand_dims(tensors[k], axis=0) for k in keys}
# for testing
# from datasets.Util.Normalization import unnormalize
# summ0 = tf.summary.image("inputs", unnormalize(tensors["inputs"][:, :, :, :3]))
# summ1 = tf.summary.image("labels", tensors["labels"] * 255)
# summ2 = tf.summary.image("old_labels", tensors["inputs"][:, :, :, 3:4])
# flow0 = tensors["inputs"][:, :, :, 4:5]
# flow0 -= tf.reduce_min(flow0)
# flow0 /= tf.reduce_max(flow0)
# flow1 = tensors["inputs"][:, :, :, 5:6]
# flow1 -= tf.reduce_min(flow1)
# flow1 /= tf.reduce_max(flow1)
# summ3 = tf.summary.image("flow_future_3", flow3)
# summ4 = tf.summary.image("flow_future_4", flow4)
# self.summaries += [summ0, summ1, summ2, summ3, summ4]
else:
augmentor_strs = self.config.unicode_list("augmentors_val", [])
assert "scale" not in augmentor_strs, "scale augmentation during test time not implemented yet"
assert "translation" not in augmentor_strs, "translation augmentation during test time not implemented yet"
augmentors = parse_augmentors(augmentor_strs, self.void_label())
tensors = [tensors for _ in range(batch_size)]
tensors = [apply_augmentors(t, augmentors) for t in tensors]
tensors = [assemble_input_tensors(t) for t in tensors]
# batchify
keys = list(tensors[0].keys())
tensors = {
k: tf.stack([t[k] for t in tensors], axis=0)
for k in keys
}
return tensors
def __init__(self, config, subset, coord):
super(CUHK03Dataset, self).__init__(subset)
assert subset in ("train", "valid"), subset
self.ignore_classes = None
self.config = config
self.subset = subset
self.coord = coord
self.data_dir = config.unicode("data_dir", CUHK03_DEFAULT_PATH)
self.model = config.unicode("model", "")
self.train_folder = config.unicode("train_folder", "train/")
self.epoch_length = config.int("epoch_length", 1000)
self.n_classes = config.int("num_classes", None)
self.input_size = config.int_list("input_size", DEFAULT_INPUT_SIZE)
self.input_size = tuple(self.input_size)
self.batching_mode = config.unicode("batching_mode", "single")
assert self.batching_mode in ("single", "pair",
"group"), self.batching_mode
self.validation_mode = config.unicode("validation_mode", "embedding")
assert self.validation_mode in ("embedding",
"similarity"), self.validation_mode
self.group_size = config.int("group_size", 4)
self.pair_ratio = config.float("pair_ratio", 1.0)
augmentor_strings = self.config.unicode_list("augmentors_train", [])
self.augmentors = parse_augmentors(augmentor_strings,
self.void_label())
self.train_names = sorted(os.listdir(self.data_dir +
self.train_folder))
train_val = numpy.array([(int(r.split('_')[0]),
int(r.split('_')[1].split('.')[0]))
for r in self.train_names])
train_id_list, train_counts = numpy.unique(train_val[:, 0],
return_counts=True)
self.train_counts = tf.constant(train_counts.astype(numpy.int32))
self.num_train_id = train_id_list.shape[0]
self.num_test_id = 6
self.idx_placeholder = tf.placeholder(tf.int32, (4, ), "idx")
self.test_case = tf.placeholder(tf.string)
self.use_end_network = tf.placeholder(tf.bool)