CodeCamp #106: Support SOLOv2 deployment with OpenVINO

[Li-Hongda]

Motivation

支持SOLOv2 mmdet3.x Openvino导出

Modification

支持SOLOv2 mmdet3.x Openvino导出

BC-Breaking(Optional)

None

Use cases (Optional)

python tools/deploy.py configs/mmdet/instance-seg/instance-seg_openvino_dynamic-800x1344.py \ 
                                       ${MMDETECTION_DIR}/configs/solo/solov2_r50_fpn_1x_coco.py \
                                       ${MMDETECTION_DIR}/demo/demo.jpg \
                                       --work-dir work_dirs/solov2/instance-seg_openvino_dynamic-800x1344 \
                                       --device cpu

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness.
3. If the modification has a dependency on downstream projects of a newer version, this PR should be tested with all supported versions of downstream projects.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[CLAassistant]

All committers have signed the CLA.

[grimoire]

Note that we have refactor rewriter in #1483 to support MMRazor.
ctx of the rewriter arguments has been removed. You can get ctx by:

ctx = FUNCTION_REWRITER.get_context()

[hanrui1sensetime]

Please solve conflicts first

[Li-Hongda]

@RunningLeon Hi, I follow your advice to add End2EndModelSOLO with this config. The latest implementation discardes bboxes and only keeps masks, scores, labels.The inference result looks right.However, maybe I didn't understand what you meant. I found this will lead to errors in the evaluation due to this line.
I also provide another simple implementation by just setting all the bboxes to zero.The result turns out to be true, but maybe not what you want.

[RunningLeon]

@Li-Hongda Hi, sorry for the confusion. Maybe a simple solution is to revert code to commit f5b3e26 and pass model_cfg in here and zero out the result.bboxes when self.model_cfg.model.type in ['SOLO', 'SOLOv2'] above here

[codecov]

Codecov Report

Patch coverage: 92.64% and project coverage change: +0.21 🎉

Comparison is base (2c7e91b) 49.44% compared to head (c3ea756) 49.65%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1521      +/-   ##
==========================================
+ Coverage   49.44%   49.65%   +0.21%     
==========================================
  Files         338      339       +1     
  Lines       12920    12985      +65     
  Branches     1897     1901       +4     
==========================================
+ Hits         6388     6448      +60     
- Misses       6088     6091       +3     
- Partials      444      446       +2     

Flag	Coverage Δ
unittests	49.65% <92.64%> (+0.21%)	⬆️

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

Impacted Files	Coverage Δ
mmdeploy/pytorch/functions/linspace.py	100.00% <ø> (ø)
...oy/codebase/mmdet/deploy/object_detection_model.py	40.89% <50.00%> (+0.09%)	⬆️
...y/codebase/mmdet/models/dense_heads/solov2_head.py	94.91% <94.91%> (ø)
...ploy/codebase/mmdet/models/dense_heads/__init__.py	100.00% <100.00%> (ø)
...loy/codebase/mmdet/models/dense_heads/solo_head.py	93.61% <100.00%> (ø)
mmdeploy/utils/test.py	87.25% <100.00%> (+0.06%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[RunningLeon]

@Li-Hongda Hi, could sync with latest main branch to fix ci error?

[RunningLeon]

@Li-Hongda Could you sync with main branch to fix ci errors? This PR is ready to merge.

[Li-Hongda]

@RunningLeon Sorry, I went on a business trip the last few days.

[RunningLeon]

LGTM