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

A Tiny Bug in agora data convertor #67

Closed
iceload123 opened this issue Jan 6, 2022 · 5 comments · Fixed by #69
Closed

A Tiny Bug in agora data convertor #67

iceload123 opened this issue Jan 6, 2022 · 5 comments · Fixed by #69
Assignees
@pangyyyyy

Comments

@iceload123
Copy link

In https://github.com/open-mmlab/mmhuman3d/blob/main/mmhuman3d/data/data_converters/agora.py#L108, it is expected to add resolution endfix to image path when res = (1280,720). However, "replace" is not a in-place operation....
@iceload123
Copy link
Author

Another small bug:
https://github.com/open-mmlab/mmhuman3d/blob/main/mmhuman3d/data/data_converters/agora.py#L121

It should be (1280,720)

@iceload123
Copy link
Author

I also find that, for posetrack dataset,
https://github.com/open-mmlab/mmhuman3d/blob/main/mmhuman3d/data/data_converters/posetrack.py#L53

assumes that "images" and "annotations" are aligned, which is not true. There is a shift as "annotations" only contains labeled frames

@iceload123
Copy link
Author

https://github.com/open-mmlab/mmhuman3d/blob/main/mmhuman3d/data/data_converters/posetrack.py#L53

I re-check the json format. "annotations" are instance-wise annotations while "images" are image-wise metas.

@iceload123
Copy link
Author

Another small bug in Surreal dataset:
dump image index starts from 1, while image path index starts from 0

https://github.com/open-mmlab/mmhuman3d/blob/main/mmhuman3d/data/data_converters/surreal.py#L187

@pangyyyyy
Copy link
Collaborator

@iceload123 Thanks for pointing out the bugs! These will be fixed shortly

@pangyyyyy pangyyyyy mentioned this issue Jan 7, 2022
Merged
caizhongang pushed a commit that referenced this issue Jan 20, 2022
@pangyyyyy
[Fix] Bug fixes in data converters and transforms (#69)
b84d9a0 
Resolves #67 
- Inplace renaming operation to obtain correct image path in agora converter
- Use tuple instead of string check in agora converter
- Retrieve correct image path using id in posetrack converter
- Fix inconsistency is cv2.inwrite image path and stored image path in surreal converter
- Fix H36M cam_param to store a dictionary of camera parameters

Resolves #70 
- Fix type in instances where `bbox_xywh` is `bbox_xyxy`
- Split `_bbox_expand` to `bbox_expand` (only bbox expansion) and `xyxy2xywh` (xyxy conversion to xywh)

Resolves #79 issue of `keypoints2d_loss` not converging
- Make a copy of keypoints2d instead of directly modifying when applying augmentations
ttxskk pushed a commit that referenced this issue Jun 17, 2022
@pangyyyyy
[Fix] Bug fixes in data converters and transforms (#69)
a4b162f 
Resolves #67 
- Inplace renaming operation to obtain correct image path in agora converter
- Use tuple instead of string check in agora converter
- Retrieve correct image path using id in posetrack converter
- Fix inconsistency is cv2.inwrite image path and stored image path in surreal converter
- Fix H36M cam_param to store a dictionary of camera parameters

Resolves #70 
- Fix type in instances where `bbox_xywh` is `bbox_xyxy`
- Split `_bbox_expand` to `bbox_expand` (only bbox expansion) and `xyxy2xywh` (xyxy conversion to xywh)

Resolves #79 issue of `keypoints2d_loss` not converging
- Make a copy of keypoints2d instead of directly modifying when applying augmentations
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@pangyyyyy pangyyyyy

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

[Fix] Bug fixes in data converters and transforms pangyyyyy/mmhuman3d
2 participants
@iceload123 @pangyyyyy