Skip to content

Jm/hotfix - #2

Merged
parkjinman98 merged 4 commits into
mainfrom
jm/hotfix
Aug 20, 2026
Merged

Jm/hotfix#2
parkjinman98 merged 4 commits into
mainfrom
jm/hotfix

Conversation

@parkjinman98

Copy link
Copy Markdown
Contributor

No description provided.

@parkjinman98 parkjinman98 self-assigned this Aug 20, 2026

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c59a7fe4dd

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread mblt_vision/utils/datasets/readiness.py
Comment thread mblt_vision/utils/datasets/readiness.py Outdated
Comment thread mblt_vision/utils/datasets/readiness.py Outdated
Comment thread mblt_vision/utils/datasets/readiness.py Outdated
Comment thread mblt_vision/utils/datasets/readiness.py
@parkjinman98

Copy link
Copy Markdown
Contributor Author

@codex review

Addressed in cae7110:

  • preserve the non-crowd COCO pose area guard while accepting the official crowd discrepancy;
  • validate WiderFace box dtype before checking the no-face sentinel;
  • normalize the all-zero WiderFace no-face sentinel before scoring;
  • retain the normalized-label fallback for DOTAv1 caches.

Not a review target: the suggestion to reject every nonpositive non-sentinel WiderFace box dimension. The supplied official WIDERFace validation metadata contains non-sentinel zero/negative-size rows, so enforcing that rule conflicts with the approved readiness compatibility direction and would reject the official dataset again.

Verification: 66 focused tests passed; official COCO pose and WIDERFace readiness both pass.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: cae7110b1a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread mblt_vision/utils/evaluation/eval_widerface.py
Comment thread mblt_vision/utils/datasets/readiness.py Outdated
@parkjinman98

Copy link
Copy Markdown
Contributor Author

@codex review

Addressed the two newly opened review threads in 8fd88c1:

  • COCO polygon segmentation now requires rasterized foreground for the union of components, preserving valid official thin components while rejecting an entirely empty mask.
  • WiderFace predictions on official no-face sentinel images now count as false positives.

Not a review target (unchanged): the earlier request to reject every non-positive-width/height WiderFace box conflicts with the official WiderFace metadata used by the organizer, which contains non-sentinel zero/negative dimension records. Readiness intentionally remains compatible with that official data.

@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review: Didn't find any major issues. 👍

Reviewed commit: 8fd88c1bf0

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

@parkjinman98
parkjinman98 merged commit aac9f54 into main Aug 20, 2026
3 checks passed
@parkjinman98
parkjinman98 deleted the jm/hotfix branch August 20, 2026 06:12
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant