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

replace skimage affine ellipse with numpy ops #1922

Merged
merged 3 commits into from
Sep 18, 2024
Merged

replace skimage affine ellipse with numpy ops #1922

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
3df8462
dropping skimage dependency in bbox affine transformation
momincks Sep 12, 2024
e8f2518
handle zero division
momincks Sep 12, 2024
453e66b
add info
momincks Sep 12, 2024
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
18 changes: 17 additions & 1 deletion albumentations/augmentations/geometric/functional.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -887,7 +887,23 @@ def bboxes_affine_ellipse(bboxes: np.ndarray, matrix: skimage.transform.Projecti
points = np.stack([x, y], axis=-1).reshape(-1, 2)

# Transform all points at once
transformed_points = skimage.transform.matrix_transform(points, matrix.params)
# Replacing skimage.transform.matrix_transform with numpy ops:
# points reshape from N, 2 to N, 3 with extra dim filled with 1
points = np.concatenate([points, np.ones((points.shape[0], 1))], axis=1)

# change matrix.params.T to matrix.T if matrix is np.ndarray and no longer skimage.transform.ProjectiveTransform
momincks marked this conversation as resolved.
Show resolved Hide resolved
transformed_points = points @ matrix.params.T
momincks marked this conversation as resolved.
Show resolved Hide resolved

# set zero to very small number before homogeneous divide
transformed_points[:, -1:] = np.where(
transformed_points[:, -1:] == 0,
np.finfo(float).eps,
transformed_points[:, -1:],
)

# homogeneous divide and then get x, y
transformed_points = (transformed_points / transformed_points[:, -1:])[:, :2]

transformed_points = transformed_points.reshape(len(bboxes), -1, 2)

# Compute new bounding boxes
Expand Down
Loading