8000 [Feature] New Codec for IPR, a wrapper for multi task losses by Tau-J · Pull Request #1628 · open-mmlab/mmpose · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

[Feature] New Codec for IPR, a wrapper for multi task losses #1628

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
Merged
merged 59 commits into from
Sep 13, 2022
Merged

[Feature] New Codec for IPR, a wrapper for multi task losses #1628

merged 59 commits into from
Sep 13, 2022

Conversation

Tau-J
Copy link
Collaborator
@Tau-J Tau-J commented Sep 2, 2022 •
edited
Loading

Motivation

  • implement a new Codec for IPR method
  • rename original IPR as SoftArgmaxHead to perform saparate coord regression
  • implement a wrapper for multi-task losses, which supports arbitrary number of losses
  • implement a new IPR head

Modification

BC-breaking (Optional)

Use cases (Optional)

Checklist

Before PR:

  • I have read and followed the workflow indicated in the CONTRIBUTING.md to create this PR.
  • Pre-commit or linting tools indicated in CONTRIBUTING.md are used to fix the potential lint issues.
  • Bug fixes are covered by unit tests, the case that causes the bug should be added in the unit tests.
  • New functionalities are covered by complete unit tests. If not, please add more unit tests to ensure correctness.
  • The documentation has been modified accordingly, including docstring or example tutorials.

After PR:

  • CLA has been signed and all committers have signed the CLA in this PR.

@Tau-J Tau-J requested review from ly015 and liqikai9 September 2, 2022 10:10
@Tau-J
Copy link
Collaborator Author
Tau-J commented Sep 2, 2022

I would like to use multi_loss_factory, but it is already used by the bottom-up method

@ly015
Copy link
Member
ly015 commented Sep 6, 2022

As for the organization of the new configs, I suggest we consider SimCC, IPR, and DSNT as algorithms parallel to top-down heatmap (td-hm) and top-down regression (td-reg, which particularly refers to deeppose or directly regressing coordinates with a FC layer).

@Tau-J
Copy link
Collaborator Author
Tau-J commented Sep 7, 2022

As for the organization of the new configs, I suggest we consider SimCC, IPR, and DSNT as algorithms parallel to top-down heatmap (td-hm) and top-down regression (td-reg, which particularly refers to deeppose or directly regressing coordinates with a FC layer).

Should we also reorganize the heads of new methods?

ly015
ly015 reviewed Sep 13, 2022
View reviewed changes
ly015
ly015 reviewed Sep 13, 2022
View reviewed changes
jin-s13
jin-s13 reviewed Sep 13, 2022
View reviewed changes
ly015
ly015 reviewed Sep 13, 2022
View reviewed changes
@ly015 ly015 merged commit 490511e into open-mmlab:dev-1.x Sep 13, 2022
ly015 added a commit that referenced this pull request Sep 13, 2022
@Tau-J @ly015
[Feature] New Codec for IPR, a wrapper for multi task losses (#1628)
a5131dc 
Co-authored-by: Yining Li <liyining0712@gmail.com>
ly015 added a commit that referenced this pull request Oct 14, 2022
@Tau-J @ly015
[Feature] New Codec for IPR, a wrapper for multi task losses (#1628)
2894af3 
Co-authored-by: Yining Li <liyining0712@gmail.com>
ly015 added a commit that referenced this pull request Oct 14, 2022
@Tau-J @ly015
[Feature] New Codec for IPR, a wrapper for multi task losses (#1628)
629f31d 
Co-authored-by: Yining Li <liyining0712@gmail.com>
ly015 added a commit to ly015/mmpose that referenced this pull request Oct 18, 2022
@Tau-J @ly015
[Feature] New Codec for IPR, a wrapper for multi task losses (open-mm…
78fb315 
…lab#1628)


Co-authored-by: Yining Li <liyining0712@gmail.com>
shuheilocale pushed a commit to shuheilocale/mmpose that referenced this pull request May 6, 2023
@Tau-J @ly015
[Feature] New Codec for IPR, a wrapper for multi task losses (open-mm…
bb40c89 
…lab#1628)


Co-authored-by: Yining Li <liyining0712@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@jin-s13 jin-s13 jin-s13 left review comments

@ly015 ly015 ly015 approved these changes

@liqikai9 liqikai9 Awaiting requested review from liqikai9

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@Tau-J @ly015 @jin-s13
0