def __getitem__(self, index: int) -> Tuple[Any, Any]:
img = np.array(Image.open(self.images[index]).convert("RGB"))
annotation = self.parse_voc_xml(
ET_parse(self.annotations[index]).getroot())["annotation"]
bbox_classes = list(
map(
lambda x: (
int(x["bndbox"]["xmin"]),
int(x["bndbox"]["ymin"]),
int(x["bndbox"]["xmax"]),
int(x["bndbox"]["ymax"]),
x["name"],
),
annotation["object"],
))
if self.transforms is not None:
result = self.transforms(image=img, bboxes=bbox_classes)
image, bbox_classes = result["image"] / 255.0, result["bboxes"]
bboxes = np.stack([a[:4] for a in bbox_classes])
labels = [self.name2class[a[4]] for a in bbox_classes]
target = {}
target["boxes"] = torch.tensor(bboxes)
target["labels"] = torch.tensor(labels)
target["image_id"] = annotation["filename"]
target["area"] = torch.tensor(
(bboxes[:, 3] - bboxes[:, 1]) * (bboxes[:, 2] - bboxes[:, 0]))
target["iscrowd"] = torch.tensor([False] * len(bboxes))
return image, target
def __getitem__(self, index: int) -> Tuple[Any, Any]:
"""
Args:
index (int): Index
Returns:
tuple: (image, target) where target is a dictionary of the XML tree.
"""
img = Image.open(self.images[index]).convert("RGB")
target = self.parse_voc_xml(ET_parse(self.annotations[index]).getroot())
if self.transforms is not None:
img, target = self.transforms(img, target)
return img, target
def __getitem__(self, index):
"""
Args:
index (int): Index
Returns:
tuple: (image, target). target are the properties as annotations in image
"""
image_name, target_class = self._breed_images[index]
image_path = join(self.images_folder, image_name)
image = Image.open(image_path).convert('RGB')
target = self.parse_voc_xml(
ET_parse(self.annotations_folder + '/' +
self.split[index][0]).getroot())
if self.transforms is not None:
image, target = self.transforms(image, target)
return image, target
def __getitem__(self, index):
img = Image.open(self.images[index]).convert('RGB')
target = self.parse_voc_xml(
ET_parse(self.annotations[index]).getroot())
targets, labels = [], []
for t in target['annotation']['object']:
bbox = [
int(t['bndbox']['xmin']),
int(t['bndbox']['ymin']),
int(t['bndbox']['xmax']),
int(t['bndbox']['ymax'])
]
label = classes.index(t['name'])
targets.append(bbox)
labels.append(label)
labels = torch.tensor(labels)
if self.transform:
img = self.transform(img)
targets = self.transform(np.array(targets))
return img, targets, labels