Skip to content

[references] Add poly scheduler for detection training#1444

Merged
felixdittrich92 merged 4 commits intomindee:mainfrom
felixdittrich92:poly
Feb 2, 2024
Merged

[references] Add poly scheduler for detection training#1444
felixdittrich92 merged 4 commits intomindee:mainfrom
felixdittrich92:poly

Conversation

@felixdittrich92
Copy link
Collaborator

This PR:

  • Add Polynomial Scheduler for detection training (TF & PT)
  • It's often mentioned in text detection papers like DB & FAST & .. so i think it is a good idea to add it :)

Any feedback is welcome 🤗

@felixdittrich92 felixdittrich92 added type: enhancement Improvement ext: references Related to references folder framework: pytorch Related to PyTorch backend framework: tensorflow topic: text detection Related to the task of text detection labels Jan 30, 2024
@felixdittrich92 felixdittrich92 added this to the 0.8.0 milestone Jan 30, 2024
@felixdittrich92 felixdittrich92 self-assigned this Jan 30, 2024
@codecov
Copy link

codecov bot commented Jan 30, 2024

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (f5445ef) 95.81% compared to head (c996c1b) 95.80%.

Additional details and impacted files
@@            Coverage Diff             @@
##             main    #1444      +/-   ##
==========================================
- Coverage   95.81%   95.80%   -0.02%     
==========================================
  Files         162      162              
  Lines        7101     7101              
==========================================
- Hits         6804     6803       -1     
- Misses        297      298       +1     
Flag Coverage Δ
unittests 95.80% <100.00%> (-0.02%) ⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

odulcy-mindee added a commit to odulcy-mindee/doctr that referenced this pull request Feb 1, 2024
@felixdittrich92
Copy link
Collaborator Author

@odulcy-mindee I think with choices it's more clean instead of raising an exception :)
Switched the default to poly -> (After testing LinkNet / DB & FAST on 2 different detection datasets x Poly / Cosine / Onecycle -> Poly has reached the best results with mostly no drops)

Copy link
Contributor

@odulcy-mindee odulcy-mindee left a comment

Choose a reason for hiding this comment

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

Yeah, it's better with choices, thanks !

@felixdittrich92 felixdittrich92 merged commit 98cd0e9 into mindee:main Feb 2, 2024
@felixdittrich92 felixdittrich92 deleted the poly branch February 2, 2024 10:21
odulcy-mindee added a commit to odulcy-mindee/doctr that referenced this pull request Feb 7, 2024
odulcy-mindee added a commit to odulcy-mindee/doctr that referenced this pull request Feb 28, 2024
odulcy-mindee added a commit to odulcy-mindee/doctr that referenced this pull request Mar 4, 2024
odulcy-mindee added a commit to odulcy-mindee/doctr that referenced this pull request Mar 5, 2024
odulcy-mindee added a commit to odulcy-mindee/doctr that referenced this pull request Mar 11, 2024
odulcy-mindee added a commit to odulcy-mindee/doctr that referenced this pull request Mar 26, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ext: references Related to references folder framework: pytorch Related to PyTorch backend topic: text detection Related to the task of text detection type: enhancement Improvement

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants