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

[Fix] Set random seed for palette if not given. #1152

Merged
merged 4 commits into from
Dec 23, 2021
Merged

[Fix] Set random seed for palette if not given. #1152

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
77b341e
Fix colors
jbwang1997 Dec 19, 2021
efec3ff
fix comments
jbwang1997 Dec 19, 2021
fa4f9cf
Add comments
jbwang1997 Dec 20, 2021
4377d11
Add comments. Add random seed in datasets
jbwang1997 Dec 22, 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
9 changes: 9 additions & 0 deletions mmseg/datasets/custom.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -349,7 +349,16 @@ def get_palette_for_custom_classes(self, class_names, palette=None):

elif palette is None:
if self.PALETTE is None:
# Get random state before set seed, and restore
# random state later.
# It will prevent loss of randomness, as the palette
# may be different in each iteration if not specified.
# See: https://github.com/open-mmlab/mmdetection/issues/5844
state = np.random.get_state()
np.random.seed(42)
# random palette
palette = np.random.randint(0, 255, size=(len(class_names), 3))
np.random.set_state(state)
else:
palette = self.PALETTE

Expand Down
9 changes: 9 additions & 0 deletions mmseg/models/segmentors/base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -245,8 +245,17 @@ def show_result(self,
seg = result[0]
if palette is None:
if self.PALETTE is None:
# Get random state before set seed,
# and restore random state later.
# It will prevent loss of randomness, as the palette
# may be different in each iteration if not specified.
# See: https://github.com/open-mmlab/mmdetection/issues/5844
state = np.random.get_state()
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can add more comments, refer to https://github.com/open-mmlab/mmdetection/blob/master/mmdet/core/visualization/image.py#L94-L96.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also, add this line in comments “The palette may be different in each iteration if not specified.”

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also, add this line in comments “The palette may be different in each iteration if not specified.”

But if set np.random.seed(42), the random palette will be fixed in each iteration.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Of course. My point is an additional straightforward explanation in comments for this modification.

np.random.seed(42)
# random palette
palette = np.random.randint(
0, 255, size=(len(self.CLASSES), 3))
np.random.set_state(state)
else:
palette = self.PALETTE
palette = np.array(palette)
Expand Down