Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

yolo bbox format #916

Closed
wants to merge 4 commits into from
Closed

yolo bbox format #916

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
8f98987
fix incorrect roundings in yolo format
May 28, 2021
d3bc5af
add test with different image sizes
May 28, 2021
844803f
Merge branch 'master' into master
Gorogorov Jun 1, 2021
9aecf57
Merge branch 'master' into master
Dipet Jun 17, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
56 changes: 31 additions & 25 deletions albumentations/augmentations/bbox_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -225,28 +225,32 @@ def convert_bbox_to_albumentations(bbox, source_format, rows, cols, check_validi
if isinstance(bbox, np.ndarray):
bbox = bbox.tolist()

if source_format == "coco":
(x_min, y_min, width, height), tail = bbox[:4], tuple(bbox[4:])
x_max = x_min + width
y_max = y_min + height
elif source_format == "yolo":
# https://github.com/pjreddie/darknet/blob/f6d861736038da22c9eb0739dca84003c5a5e275/scripts/voc_label.py#L12
if source_format == "yolo":
bbox, tail = bbox[:4], tuple(bbox[4:])
_bbox = np.array(bbox[:4])
if np.any((_bbox <= 0) | (_bbox > 1)):
raise ValueError("In YOLO format all labels must be float and in range (0, 1]")

x, y, width, height = denormalize_bbox(bbox, rows, cols)
x, y, width, height = bbox
dx, dy = 1 / rows, 1 / cols

x_min = x - width / 2 + dx
x_max = x_min + width
y_min = y - height / 2 + dy
y_max = y_min + height

x_min = int(x - width / 2 + 1)
x_max = int(x_min + width)
y_min = int(y - height / 2 + 1)
y_max = int(y_min + height)
bbox = (x_min, y_min, x_max, y_max) + tail
else:
(x_min, y_min, x_max, y_max), tail = bbox[:4], tuple(bbox[4:])
if source_format == "coco":
(x_min, y_min, width, height), tail = bbox[:4], tuple(bbox[4:])
x_max = x_min + width
y_max = y_min + height
else:
(x_min, y_min, x_max, y_max), tail = bbox[:4], tuple(bbox[4:])

bbox = (x_min, y_min, x_max, y_max) + tail
bbox = normalize_bbox(bbox, rows, cols)

bbox = (x_min, y_min, x_max, y_max) + tail
bbox = normalize_bbox(bbox, rows, cols)
if check_validity:
check_bbox(bbox)
return bbox
Expand Down Expand Up @@ -280,20 +284,22 @@ def convert_bbox_from_albumentations(bbox, target_format, rows, cols, check_vali
)
if check_validity:
check_bbox(bbox)
bbox = denormalize_bbox(bbox, rows, cols)
if target_format == "coco":
(x_min, y_min, x_max, y_max), tail = bbox[:4], tuple(bbox[4:])
width = x_max - x_min
height = y_max - y_min
bbox = (x_min, y_min, width, height) + tail
elif target_format == "yolo":
# https://github.com/pjreddie/darknet/blob/f6d861736038da22c9eb0739dca84003c5a5e275/scripts/voc_label.py#L12
if target_format == "yolo":
(x_min, y_min, x_max, y_max), tail = bbox[:4], bbox[4:]
x = int((x_min + x_max) / 2 - 1)
y = int((y_min + y_max) / 2 - 1)
dx, dy = 1 / rows, 1 / cols
x = (x_min + x_max) / 2 - dx
y = (y_min + y_max) / 2 - dy
width = x_max - x_min
height = y_max - y_min
bbox = normalize_bbox((x, y, width, height) + tail, rows, cols)
bbox = (x, y, width, height) + tail
else:
bbox = denormalize_bbox(bbox, rows, cols)
if target_format == "coco":
(x_min, y_min, x_max, y_max), tail = bbox[:4], tuple(bbox[4:])
width = x_max - x_min
height = y_max - y_min
bbox = (x_min, y_min, width, height) + tail

return bbox


Expand Down
20 changes: 11 additions & 9 deletions tests/test_bbox.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,7 @@ def test_convert_bbox_to_albumentations(bbox, source_format, expected):
converted_bbox = convert_bbox_to_albumentations(
bbox, rows=image.shape[0], cols=image.shape[1], source_format=source_format
)
assert converted_bbox == expected
assert np.all(np.isclose(converted_bbox, expected))


@pytest.mark.parametrize(
Expand Down Expand Up @@ -130,14 +130,16 @@ def test_convert_bbox_from_albumentations(bbox, target_format, expected):
],
)
def test_convert_bbox_to_albumentations_and_back(bbox, bbox_format):
image = np.ones((100, 100, 3))
converted_bbox = convert_bbox_to_albumentations(
bbox, rows=image.shape[0], cols=image.shape[1], source_format=bbox_format
)
converted_back_bbox = convert_bbox_from_albumentations(
converted_bbox, rows=image.shape[0], cols=image.shape[1], target_format=bbox_format
)
assert np.all(np.isclose(converted_back_bbox, bbox))
for rows in range(100, 105):
for cols in range(100, 105):
image = np.ones((cols, rows, 3))
converted_bbox = convert_bbox_to_albumentations(
bbox, rows=image.shape[0], cols=image.shape[1], source_format=bbox_format
)
converted_back_bbox = convert_bbox_from_albumentations(
converted_bbox, rows=image.shape[0], cols=image.shape[1], target_format=bbox_format
)
assert np.all(np.isclose(converted_back_bbox, bbox))


def test_convert_bboxes_to_albumentations():
Expand Down