def _get_metainfo(self, index):
question = gdef.translate_question(self.questions[index])
scene = gdef.translate_scene(self.scenes[question['image_index']])
question['scene'] = scene
question['image_index'] = question['image_index']
question['image_filename'] = gdef.get_image_filename(scene)
question['question_index'] = index
question['question_tokenized'] = nltk.word_tokenize(question['question'])
# program section
has_program = False
if 'program_nsclseq' in question:
question['program_raw'] = question['program_nsclseq']
question['program_seq'] = question['program_nsclseq']
has_program = True
elif 'program' in question:
question['program_raw'] = question['program']
question['program_seq'] = gdef.program_to_nsclseq(question['program'], question)
has_program = True
if has_program:
question['program_tree'] = nsclseq_to_nscltree(question['program_seq'])
question['program_qsseq'] = nsclseq_to_nsclqsseq(question['program_seq'])
question['program_qstree'] = nscltree_to_nsclqstree(question['program_tree'])
question['question_type'] = question['program_seq'][-1]['op']
else:
question['question_type'] = None
return question
def _get_metainfo(self, index):
question = gdef.translate_question(self.questions[index])
scene = gdef.translate_scene(self.scenes[question["image_index"]])
question["scene"] = scene
question["image_index"] = question["image_index"]
question["image_filename"] = gdef.get_image_filename(scene)
question["question_index"] = index
question["question_tokenized"] = nltk.word_tokenize(question["question"])
# program section
has_program = False
if "program_nsclseq" in question:
question["program_raw"] = question["program_nsclseq"]
question["program_seq"] = question["program_nsclseq"]
has_program = True
elif "program" in question:
question["program_raw"] = question["program"]
question["program_seq"] = gdef.program_to_nsclseq(
question["program"], question
)
has_program = True
if has_program:
question["program_tree"] = nsclseq_to_nscltree(question["program_seq"])
question["program_qsseq"] = nsclseq_to_nsclqsseq(question["program_seq"])
question["program_qstree"] = nscltree_to_nsclqstree(
question["program_tree"]
)
question["question_type"] = question["program_seq"][-1]["op"]
else:
question["question_type"] = None
return question
def _get_metainfo(self, index):
question = dict()
question['scene'] = gdef.translate_scene(self.scenes[index])
question['image_index'] = index
question['image_filename'] = gdef.get_image_filename(question['scene'])
# program section
question['program_raw'] = self.program
question['program_seq'] = self.program
question['program_tree'] = nsclseq_to_nscltree(question['program_seq'])
question['program_qsseq'] = nsclseq_to_nsclqsseq(question['program_seq'])
question['program_qstree'] = nscltree_to_nsclqstree(question['program_tree'])
question['question_type'] = question['program_seq'][-1]['op']
return question
def _get_metainfo(self, index):
question = dict()
question["scene"] = gdef.translate_scene(self.scenes[index])
question["image_index"] = index
question["image_filename"] = gdef.get_image_filename(question["scene"])
# program section
question["program_raw"] = self.program
question["program_seq"] = self.program
question["program_tree"] = nsclseq_to_nscltree(question["program_seq"])
question["program_qsseq"] = nsclseq_to_nsclqsseq(question["program_seq"])
question["program_qstree"] = nscltree_to_nsclqstree(question["program_tree"])
question["question_type"] = question["program_seq"][-1]["op"]
return question
def _get_metainfo(self, index):
qid = index // len(VIEWS)
view_id = index % len(VIEWS)
question = gdef.translate_question(self.questions[qid])
question = copydict(question)
question["index"] = index
scene = gdef.translate_scene(self.scenes[question["image_index"]])
scene = copydict(scene)
scene["objects_detection"] = scene["objects_detection"][VIEWS[view_id]]
for obj in scene["objects"]:
obj["mask"] = obj["mask"][VIEWS[view_id]]
question["scene"] = scene
# question["image_index"] = question["image_index"]
question["view_id"] = view_id
question["image_filename"] = osp.join(
gdef.get_image_filename(scene), VIEWS[view_id] + ".png"
)
question["question_index"] = qid
question["question_tokenized"] = nltk.word_tokenize(question["question"])
# program section
has_program = False
if "program_nsclseq" in question:
question["program_raw"] = question["program_nsclseq"]
question["program_seq"] = question["program_nsclseq"]
has_program = True
elif "program" in question:
question["program_raw"] = question["program"]
question["program_seq"] = gdef.program_to_nsclseq(
question["program"], question
)
has_program = True
if has_program:
question["program_tree"] = nsclseq_to_nscltree(question["program_seq"])
question["program_qsseq"] = nsclseq_to_nsclqsseq(question["program_seq"])
question["program_qstree"] = nscltree_to_nsclqstree(
question["program_tree"]
)
question["question_type"] = question["program_seq"][-1]["op"]
else:
question["question_type"] = None
return question
def __getitem__(self, index):
feed_dict = deepcopy(self.get_metainfo(index))
image_index = feed_dict['image_index']
if self.image_root is not None:
feed_dict['image'] = Image.open(
osp.join(self.image_root,
feed_dict['image_filename'])).convert('RGB')
feed_dict['image'] = self.image_transform(feed_dict['image'])
if self.features_h5 is not None:
feed_dict['feature'] = self.features[image_index]
feed_dict['question_raw'] = feed_dict.pop('question')
feed_dict['question_raw_tokenized'] = feed_dict['question_raw'].split()
feed_dict['question_type'] = 'exist'
feed_dict['program_seq'] = gdef.program_to_nsclseq(
feed_dict.pop('program'))
feed_dict['program_qsseq'] = nsclseq_to_nsclqsseq(
feed_dict['program_seq'])
feed_dict['answer'] = gdef.canonize_answer(feed_dict['answer'],
feed_dict['question_type'])
return feed_dict
