def test_can_save_and_load_segmentation_with_unordered_labels(self):
source_label_map = {
'background': (0, 0, 0),
'label_1': (10, 10, 10),
'label_0': (20, 20, 20)
}
source_dataset = Dataset.from_iterable([
DatasetItem(id='a', image=np.ones((1, 5, 3)),
annotations=[
Mask(image=np.array([[1, 0, 0, 0, 0]]),
attributes={'is_crowd': False}, label=0, id=1),
Mask(image=np.array([[0, 1, 0, 0, 0]]),
attributes={'is_crowd': False}, label=1, id=1),
Mask(image=np.array([[0, 0, 1, 1, 1]]),
attributes={'is_crowd': False}, label=2, id=2)
]
)
], categories=make_kitti_categories(source_label_map))
expected_label_map = {
'background': (0, 0, 0),
'label_0': (20, 20, 20),
'label_1': (10, 10, 10)
}
expected_dataset = Dataset.from_iterable([
DatasetItem(id='a', image=np.ones((1, 5, 3)),
annotations=[
Mask(image=np.array([[1, 0, 0, 0, 0]]),
attributes={'is_crowd': False}, label=0, id=1),
Mask(image=np.array([[0, 1, 0, 0, 0]]),
attributes={'is_crowd': False}, label=2, id=1),
Mask(image=np.array([[0, 0, 1, 1, 1]]),
attributes={'is_crowd': False}, label=1, id=2)
]
)
], categories=make_kitti_categories(expected_label_map))
with TestDir() as test_dir:
self._test_save_and_load(source_dataset,
partial(KittiConverter.convert, tasks=KittiTask.segmentation,
label_map=source_label_map), test_dir,
target_dataset=expected_dataset)
def test_can_write_and_parse_dataset_meta_file(self):
src_label_map = KittiLabelMap
with TestDir() as test_dir:
source_dataset = Dataset.from_iterable([],
categories=make_kitti_categories(src_label_map))
KittiConverter.convert(source_dataset, test_dir,
save_dataset_meta=True)
dst_label_map = parse_meta_file(test_dir)
self.assertEqual(src_label_map, dst_label_map)
def categories(self):
label_map = OrderedDict()
label_map['background'] = None
label_map['Label_1'] = None
label_map['label_2'] = None
return make_kitti_categories(label_map)
def categories(self):
label_map = OrderedDict()
label_map['a'] = None
label_map['b'] = None
return make_kitti_categories(label_map)
def categories(self):
return make_kitti_categories()