Add a unified resize operation for detection and a resize op for recognition inference #295
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
SamitHuang
requested review from
hadipash,
zhtmike,
HaoyangLee and
Songyuanwei
and removed request for
hadipash
SamitHuang
changed the title
hadipash
approved these changes
May 19, 2023
Comment on lines
+281
to
+291
| padded_img = np.zeros((tar_h, tar_w, 3), dtype=np.uint8) | ||
| padded_img[:resize_h, :resize_w, :] = resized_img | ||
| data['image'] = padded_img |
There was a problem hiding this comment.
Suggested change
| padded_img = np.zeros((tar_h, tar_w, 3), dtype=np.uint8) | |
| padded_img[:resize_h, :resize_w, :] = resized_img | |
| data['image'] = padded_img | |
| data['image'] = np.pad(data['image'], ((0, tar_h - resize_h), (0, tar_w - resize_w), (0, 0))) |
Comment on lines
+290
to
+301
| data['polys'][:, :, 0] = data['polys'][:, :, 0] * scale_w | ||
| data['polys'][:, :, 1] = data['polys'][:, :, 1] * scale_h |
There was a problem hiding this comment.
Suggested change
| data['polys'][:, :, 0] = data['polys'][:, :, 0] * scale_w | |
| data['polys'][:, :, 1] = data['polys'][:, :, 1] * scale_h | |
| data['polys'] = data['polys'] * [scale_w, scale_h] |
| resize_h = self.tar_h | ||
|
|
||
| if self.keep_ratio==False: | ||
| assert self.tar_w is not None, 'Must specify target_width if keep_ratio is False' |
There was a problem hiding this comment.
Move assert inside __init__?
Comment on lines
+362
to
+364
| padded_img = np.zeros((self.tar_h, self.tar_w, 3), dtype=np.uint8) | ||
| padded_img[:, :resize_w, :] = resized_img | ||
| data['image'] = padded_img |
There was a problem hiding this comment.
Suggested change
| padded_img = np.zeros((self.tar_h, self.tar_w, 3), dtype=np.uint8) | |
| padded_img[:, :resize_w, :] = resized_img | |
| data['image'] = padded_img | |
| data['image'] = np.pad(data['image'], ((0, 0), (0, self.tar_w - resize_w), (0, 0))) |
| - DetResize: | ||
| target_size: [ 1152, 2048] | ||
| keep_ratio: True | ||
| limit_type: auto |
There was a problem hiding this comment.
I feel this is a bit confusing. When I want an image to have a specific resolution (set by target_size) and keep its ratio, I must need to set limit_type to auto. Otherwise, I will get completely unexpected output.
There was a problem hiding this comment.
Don't have to set limit_type auto in this case. You can just use the default "min". Here auto is just to make the same as ScalePadImage.
There was a problem hiding this comment.
If I don't set limit_type to auto, the output will be of size 736 by the shortest side.
HaoyangLee
approved these changes
May 19, 2023
SamitHuang
changed the title
|
under further improvment. |
SamitHuang
force-pushed
the
base
branch
2 times, most recently
from
32ebb72 to
d1cd911
Compare
SamitHuang
changed the title
Songyuanwei
approved these changes
May 24, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Thank you for your contribution to the MindOCR repo.
Before submitting this PR, please make sure:
Motivation
(Write your motivation for proposed changes here.)
Test Plan
(How should this PR be tested? Do you require special setup to run the test or repro the fixed bug?)
Related Issues and PRs
(Is this PR part of a group of changes? Link the other relevant PRs and Issues here. Use https://help.github.com/en/articles/closing-issues-using-keywords for help on GitHub syntax)