def read_file_from_db(self, is_training: bool, train_key: List,
val_key: List) -> str:
if is_training:
if train_key is None:
raise Exception(
"Initialized with validation only. Training data not downloaded."
)
return DATA_STORE.get_file(train_key[0])["fpath"]
return DATA_STORE.get_file(val_key[0])["fpath"]
def read_file_from_db(self, is_train, task_key, sample=True):
if sample and not is_train:
task_path = task_key + "_test"
else:
task_path = task_key + "_train" if is_train else task_key
if not DATA_STORE.is_valid(task_path):
raise NameError("{0} does not exist.".format(task_path))
return DATA_STORE.get_file(task_path)["fpath"]
def val_data_stream(self):
if self.images_val is None:
raise Exception(
"Download val_data first by setting validation_only to True")
for img_id, q_id, a_id in self.index:
img_key = "vqa/images-val/val2014/COCO_val2014_%012d" % img_id
img_file = DATA_STORE.get_file(img_key)
img = self.image_from_file(img_file["fpath"])
q = self.question_index[q_id]
#TODO: Fix this hardcode
a = self.answer_index(True)[(q_id, a_id)]
yield (img, q, a)
def train_data_stream(self):
if self.images_train is None:
raise Exception(
"Download train_data_first by setting validation_only to False"
)
if not self.is_training:
raise Exception(
"Build index first with build_index(is_training=True)")
for img_id, q_id, a_id in self.index:
img_key = "vqa/images-train/train2014/COCO_train2014_%012d" % img_id
img_file = DATA_STORE.get_file(img_key)
img = self.image_from_file(img_file["fpath"])
q = self.question_index[q_id]
#TODO: Fix this hardcode
a = self.answer_index(True)[(q_id, a_id)]
yield (img, q, a)
return