Make OHEM work with seesaw loss

[ohwi]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Since seesaw loss's return_dict does not contain the key loss_cls, OHEMSampler does not work with seesaw loss.

Modification

Make the key for return_dict to be controllable.

BC-breaking (Optional)

By making default key for return_dict be loss_cls, users only need to change the key for return_dict for some specific losses like seesaw loss.

Use cases (Optional)

Specify the key for loss's return_dict like below:

# Using seesaw loss
sampler=dict(
    type='OHEMSampler',
    num=512,
    pos_fraction=0.25,
    neg_pos_ub=-1,
    add_gt_as_proposals=True,
    loss_key='loss_cls_objectness',
),

[CLAassistant]

All committers have signed the CLA.

[hhaAndroid]

@ohwi Thank you for your contribution.

[codecov]

Codecov Report

Merging #6514 (eae609a) into master (c76ab0e) will decrease coverage by 0.00%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #6514      +/-   ##
==========================================
- Coverage   62.04%   62.03%   -0.01%     
==========================================
  Files         319      319              
  Lines       25343    25344       +1     
  Branches     4194     4194              
==========================================
  Hits        15723    15723              
- Misses       8794     8795       +1     
  Partials      826      826              

Flag	Coverage Δ
unittests	62.01% <100.00%> (+<0.01%)	⬆️

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

Impacted Files	Coverage Δ
mmdet/core/bbox/samplers/ohem_sampler.py	84.21% <100.00%> (+0.42%)	⬆️
mmdet/models/roi_heads/mask_heads/maskiou_head.py	87.35% <0.00%> (-2.30%)	⬇️
mmdet/core/bbox/assigners/max_iou_assigner.py	73.68% <0.00%> (+1.31%)	⬆️

---

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 c76ab0e...eae609a. Read the comment docs.

[ohwi]

Hi. Sorry for asking this, I'm not familiar with code coverage. Is there something that I need to do change in the file maskiod_head.py? I haven't touched that file.

[RangiLyu]

Hi. Sorry for asking this, I'm not familiar with code coverage. Is there something that I need to do change in the file maskiod_head.py? I haven't touched that file.

Don't worry about that.

[OpenMMLab-Coodinator]

Hi @ohwi！First of all, we want to express our gratitude for your significant PR in the MMDet project. Your contribution is highly appreciated, and we are grateful for your efforts in helping improve this open-source project during your personal time. We believe that many developers will benefit from your PR.

We would also like to invite you to join our Special Interest Group (SIG) private channel on Discord, where you can share your experiences, ideas, and build connections with like-minded peers. To join the SIG channel, simply message moderator— OpenMMLab on Discord or briefly share your open-source contributions in the #introductions channel and we will assist you. Look forward to seeing you there! Join us ：https://discord.gg/raweFPmdzG

If you are Chinese or have WeChat，welcome to join our community on WeChat. You can add our assistant ：openmmlabwx. Please add "mmsig + Github ID" as a remark when adding friends：）
Thank you again for your contribution❤