in src/pixparse/data/datasets_utils.py [0:0]
def __getitem__(self, index):
if self.split == "test":
entry = self.data_dict['data'][index]
labels = "<s_question>" + entry['question'] + "</s_question>"
img_path = os.path.join(self.img_dir, entry['image'])
question_id = entry['questionId']
image_id = entry["image"]
if self.split == "val":
entry = self.data_dict['data'][index]
labels = {"question": entry['question'], "answers": entry['answers']}
img_path = os.path.join(self.img_dir, entry['image'])
question_id = entry['questionId']
image_id = entry["image"]
else:
image_id = self.all_images[index]
questions_and_answers = self.data_dict[image_id]
labels = questions_and_answers
img_path = os.path.join(self.img_dir, image_id)
question_id = -1 # Not parsed from original dataset.
image = Image.open(img_path).convert("L")
if self.transform:
image = self.transform(image)
return {"image": image, "labels": labels, "image_id": image_id, "question_id": question_id}