Add Semi-SL for Multi-class Classification (EfficientNet-B0)

[harimkang]

Summary

CVS-141546

• OTX 2.1 imports the Semi-SL algorithm from OTX 1.6.
• Add Unlabeled data pipeline

--data_root: Target the supervised folder.
--data.config.unlabeled_subset.data_root: Targets the unlabeled folder.

For the example above,
--data_root is supervised/
and --data.config.unlabeled_subset.data_root must be specified as unlabeled/.

• Add Multi-Transforms logic into unlabeled dataset
  
• Add Semi-SL classifier & Head in OTX 1.6

FMNIST 25@1_0	Accuracy	E2E time	Epochs
Baseline (SL)	79.01	0:03:37	90
Semi-SL	83.03 (+4.02)	0:06:19 (1.75x)	75

How to test

otx train --config src/otx/recipe/classification/multi_class_cls/efficientnet_b0_semisl.yaml --data_root fmnist@25_0/supervised --data.config.unlabeled_subset.data_root fmnist@25_0/unlabeled

Checklist

• I have added unit tests to cover my changes.​
• I have added integration tests to cover my changes.​
• I have ran e2e tests and there is no issues.
• I have added the description of my changes into CHANGELOG in my target branch (e.g., CHANGELOG in develop).​
• I have updated the documentation in my target branch accordingly (e.g., documentation in develop).
• I have linked related issues.

License

• I submit my code changes under the same Apache License that covers the project.
  Feel free to contact the maintainers if that's a concern.
• I have updated the license header for each file (see an example below).

# Copyright (C) 2024 Intel Corporation
# SPDX-License-Identifier: Apache-2.0

[codecov]

Codecov Report

Attention: Patch coverage is 30.23256% with 150 lines in your changes missing coverage. Please review.

Project coverage is 82.51%. Comparing base (81c2d9e) to head (42f5404).

❗ Current head 42f5404 differs from pull request most recent head fd5c7ca

Please upload reports for the commit fd5c7ca to get more accurate results.

Files	Patch %	Lines
src/otx/algo/classification/heads/semi_sl_head.py	20.23%	67 Missing ⚠️
src/otx/core/data/module.py	31.81%	30 Missing ⚠️
src/otx/algo/callbacks/unlabeled_loss_warmup.py	0.00%	19 Missing ⚠️
src/otx/algo/classification/efficientnet.py	27.27%	16 Missing ⚠️
...go/classification/classifier/semi_sl_classifier.py	31.25%	11 Missing ⚠️
src/otx/algo/callbacks/iteration_timer.py	40.00%	6 Missing ⚠️
src/otx/core/data/dataset/base.py	80.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #3566      +/-   ##
===========================================
- Coverage    82.86%   82.51%   -0.35%     
===========================================
  Files          256      258       +2     
  Lines        25305    25394      +89     
===========================================
- Hits         20968    20954      -14     
- Misses        4337     4440     +103     

Flag	Coverage Δ
py310	?
py311	82.51% <30.23%> (-0.06%)	⬇️

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

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[kprokofi]

Thanks for your work! Overall design looks good.

[kprokofi]

Can we simplify --data.config.unlabeled_subset.data_root CLI command? I prefer shorter, maybe we can create a link to this param and call it as "--unlabeled_data_root"

[harimkang]

Can we simplify --data.config.unlabeled_subset.data_root CLI command? I prefer shorter, maybe we can create a link to this param and call it as "--unlabeled_data_root"

Looking at it, the current data-related configs are gathered in data.config. For now, I think we need to think a little more about how to make this a little shorter. Personally, I think it would be better if it goes along with the refactoring of OTXDatamodule.

[sungchul2]

LGTM