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.

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

[BC-breaking] Update KeypointRCNN weights #1609

Merged
merged 3 commits into from
Dec 5, 2019

Conversation

fmassa
Copy link
Member

@fmassa fmassa commented Nov 25, 2019

Fixes #1606

We support old weights by passing pretrained='legacy', but the default behavior is to pick the new set of weights.
As such, this is a BC-breaking change.

So if users relied on the pretrained features from KeypointRCNN for training classifiers (and only saved the classifiers, picking the weight from torchvision pretrained model), they will need to change to pretrained='legacy'.

@codecov-io
Copy link

codecov-io commented Nov 25, 2019

Codecov Report

Merging #1609 into master will decrease coverage by 0.02%.
The diff coverage is 0%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master    #1609      +/-   ##
==========================================
- Coverage   65.65%   65.63%   -0.03%     
==========================================
  Files          90       90              
  Lines        7085     7088       +3     
  Branches     1075     1076       +1     
==========================================
  Hits         4652     4652              
- Misses       2126     2129       +3     
  Partials      307      307
Impacted Files Coverage Δ
torchvision/models/detection/keypoint_rcnn.py 78.26% <0%> (-3.56%) ⬇️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b590f8c...10d0858. Read the comment docs.

@fmassa fmassa merged commit 6bf920b into pytorch:master Dec 5, 2019
@fmassa fmassa deleted the update-krcnn-weights branch December 5, 2019 16:46
fmassa added a commit to fmassa/vision-1 that referenced this pull request Jan 28, 2020
* Update KeypointRCNN weights with correct file

* Fix model

* Fix
facebook-github-bot pushed a commit that referenced this pull request Jan 28, 2020
Summary:
* Update KeypointRCNN weights with correct file

* Fix model

* Fix
Pull Request resolved: #1801

Reviewed By: lerks

Differential Revision: D19598576

Pulled By: fmassa

fbshipit-source-id: ddd116b84d36cef7aace8d42397296f5a1dd5c78
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.

Cannot reproduce performance of Keypoint R-CNN with ResNet-50
3 participants