[Feature] Refactor Inferencers

[gaotongxiao]

Refactor inferencers and their entry script based on mmengine.BaseInferencer

For colleagues: BaseMMOCRInferencer, TextDetInferencer and TextRecInferencer are good reference points.

Attention

To users who have been using 1.x/dev-1.x branch on Windows and wish to upgrade MMOCR to use new Inferencers: You may encounter the following error:

FileNotFoundError: class `CRNN` in mmocr/models/textrecog/recognizers/crnn.py: class `CRNNDecoder` in mmocr/models/textrecog/decoders/crnn_decoder.py: class `Dictionary` in mmocr/models/common/dictionary/dictionary.py: [Errno 2] No such file or directory: 'd:/a/mmocr/mmocr/mmocr/.mim/configs/textrecog/crnn/../../../dicts/lower_english_digits.txt'

Which has to be manually fixed by running the command under mmocr\ root directory:

mklink /D mmocr\.mim\dicts dicts\

[codecov]

Codecov Report

Base: 85.58% // Head: 88.42% // Increases project coverage by +2.84% 🎉

Coverage data is based on head (a0291b2) compared to base (2d743cf).
Patch coverage: 84.61% of modified lines in pull request are covered.

❗ Current head a0291b2 differs from pull request most recent head 7777a52. Consider uploading reports for the commit 7777a52 to get more accurate results

Additional details and impacted files

@@             Coverage Diff             @@
##           dev-1.x    #1608      +/-   ##
===========================================
+ Coverage    85.58%   88.42%   +2.84%     
===========================================
  Files          175      175              
  Lines        10950    10881      -69     
  Branches      1526     1523       -3     
===========================================
+ Hits          9372     9622     +250     
+ Misses        1313      979     -334     
- Partials       265      280      +15     

Flag	Coverage Δ
unittests	88.42% <84.61%> (+2.84%)	⬆️

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

Impacted Files	Coverage Δ
mmocr/ocr.py	0.00% <0.00%> (ø)
mmocr/apis/inferencers/textspot_inferencer.py	35.71% <35.71%> (ø)
mmocr/apis/inferencers/textdet_inferencer.py	92.30% <80.00%> (+50.64%)	⬆️
mmocr/apis/inferencers/mmocr_inferencer.py	95.27% <88.63%> (+77.73%)	⬆️
mmocr/apis/inferencers/base_mmocr_inferencer.py	90.42% <95.00%> (+67.48%)	⬆️
mmocr/apis/inferencers/kie_inferencer.py	81.90% <97.87%> (+61.90%)	⬆️
mmocr/apis/inferencers/textrec_inferencer.py	100.00% <100.00%> (+40.00%)	⬆️
mmocr/datasets/transforms/loading.py	96.89% <100.00%> (+7.38%)	⬆️
mmocr/datasets/wildreceipt_dataset.py	100.00% <100.00%> (+4.00%)	⬆️
...cr/models/textdet/module_losses/pan_module_loss.py	89.10% <0.00%> (-7.70%)	⬇️
... and 19 more

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 at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.