Skip to content

Conversation

@xvjiarui
Copy link
Collaborator

Fixed #4322

Before #4087, score_thr is applied before multiply score_factor.
However, in #4087, the order is switched, which led to the degenerated performance for YOLOv3 and FCOS.

@codecov
Copy link

codecov bot commented Jan 19, 2021

Codecov Report

Merging #4473 (78a9dc0) into master (405c377) will increase coverage by 0.06%.
The diff coverage is 0.00%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master    #4473      +/-   ##
==========================================
+ Coverage   61.93%   62.00%   +0.06%     
==========================================
  Files         241      241              
  Lines       18357    18359       +2     
  Branches     3075     3075              
==========================================
+ Hits        11369    11383      +14     
+ Misses       6430     6418      -12     
  Partials      558      558              
Flag Coverage Δ
unittests 61.96% <0.00%> (+0.06%) ⬆️

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

Impacted Files Coverage Δ
mmdet/core/post_processing/bbox_nms.py 80.32% <0.00%> (-2.73%) ⬇️
mmdet/models/detectors/cornernet.py 94.87% <0.00%> (-5.13%) ⬇️
mmdet/models/dense_heads/corner_head.py 74.45% <0.00%> (-1.91%) ⬇️
mmdet/models/roi_heads/mask_heads/maskiou_head.py 100.00% <0.00%> (+5.37%) ⬆️
mmdet/models/roi_heads/mask_scoring_roi_head.py 87.50% <0.00%> (+32.14%) ⬆️

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 405c377...78a9dc0. Read the comment docs.

@ZwwWayne
Copy link
Collaborator

Kindly ping @RunningLeon, does the modification affect the ONNX exportation?

@ZwwWayne ZwwWayne merged commit 6031206 into open-mmlab:master Jan 24, 2021
@RunningLeon
Copy link
Collaborator

@ZwwWayne No bad impacts with PyTorch2ONNX.

@shinya7y shinya7y mentioned this pull request Apr 16, 2021
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.

APs of YOLOv3 pretrained models drop over 0.9 after updating to v2.7 and some other detectors drop 0.1

3 participants