masks from (n,h,w) to [(h,w),...,(h,w)]

[howardchina]

https://github.com/albumentations-team/albumentations/blob/master/albumentations/core/composition.py#L223
"data" is List

[CLAassistant]

All committers have signed the CLA.

[codecov]

Codecov Report

Merging #4032 (b4633be) into master (b515fb9) will increase coverage by 0.07%.
The diff coverage is 53.00%.

@@            Coverage Diff             @@
##           master    #4032      +/-   ##
==========================================
+ Coverage   61.78%   61.86%   +0.07%     
==========================================
  Files         224      225       +1     
  Lines       16632    16672      +40     
  Branches     2810     2825      +15     
==========================================
+ Hits        10276    10314      +38     
+ Misses       5859     5853       -6     
- Partials      497      505       +8     

Flag	Coverage Δ
unittests	61.82% <53.00%> (+0.07%)	⬆️

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

Impacted Files	Coverage Δ
mmdet/apis/inference.py	20.48% <0.00%> (-1.04%)	⬇️
mmdet/apis/test.py	13.51% <0.00%> (-0.13%)	⬇️
mmdet/core/anchor/anchor_generator.py	78.98% <ø> (ø)
...mdet/core/bbox/iou_calculators/iou2d_calculator.py	96.77% <ø> (ø)
mmdet/models/dense_heads/anchor_head.py	85.84% <ø> (ø)
mmdet/models/dense_heads/atss_head.py	27.94% <ø> (-0.02%)	⬇️
mmdet/models/dense_heads/gfl_head.py	27.35% <ø> (-0.04%)	⬇️
mmdet/models/dense_heads/vfnet_head.py	31.48% <ø> (+0.07%)	⬆️
mmdet/models/dense_heads/yolo_head.py	34.04% <0.00%> (-0.19%)	⬇️
mmdet/models/losses/focal_loss.py	43.90% <0.00%> (ø)
... and 22 more

---

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 b515fb9...dd615e0. Read the comment docs.

[yhcao6]

Can be simplified by np.vsplit

[hellock]

Task linked: CU-4zxm6k [Review] MMdet #4032

[yhcao6]

@howardchina, please sign CLAassistant, then we can merge.

[howardchina]

CLAassistant is signed already.

[Adblu]

Thanks for the fantastic work regarding new release !
I have 1 question regarding TODO in transforms.py , my training crushes when I'm using albu augmentation. I have found a bug fix, however its not elegant and brings lot of problems.
Are you planning to resolve it in future @howardchina @ZwwWayne @yhcao6 ?

[ZwwWayne]

Please @yhcao6 have a look for that.

[Adblu]

@ZwwWayne @yhcao6 Is there any update on the issue ?