dev(narugo): add classify timm's unittest (19c0210f) · Commits · git-mirror / Imgutils

imgutils/generic/classify_timm.py

+2 −1

Original line number	Diff line number	Diff line
		@@ -109,7 +109,8 @@ class ClassifyTIMMModel:
		elif preprocessor == 'val':
		trans = val_trans
		else:
		raise ValueError(f'Unknown processor - {preprocessor!r}.')
		raise ValueError(
		f'Unknown processor, "test" or "val" expected but {preprocessor!r} found.') # pragma: no cover

		input_ = trans(image)[None, ...]
		output_names = [output.name for output in model.get_outputs()]

imgutils/generic/multilabel_timm.py

+2 −1

Original line number	Diff line number	Diff line
		@@ -226,7 +226,8 @@ class MultiLabelTIMMModel:
		elif preprocessor == 'val':
		trans = val_trans
		else:
		raise ValueError(f'Unknown processor - {preprocessor!r}.') # pragma: no cover
		raise ValueError(
		f'Unknown processor, "test" or "val" expected but {preprocessor!r} found.') # pragma: no cover

		input_ = trans(image)[None, ...]
		output_names = [output.name for output in model.get_outputs()]

test/generic/test_classify_timm.py

0 → 100644

+111 −0

Original line number	Diff line number	Diff line
		import pytest

		from imgutils.generic.classify_timm import _open_models_for_repo_id, classify_timm_predict
		from test.testings import get_testfile


		@pytest.fixture(scope='module', autouse=True)
		def _release_model_after_run():
		try:
		yield
		finally:
		_open_models_for_repo_id.cache_clear()


		@pytest.mark.unittest
		class TestGenericClassifyTIMM:
		def test_classify_timm_predict_1(self):
		assert classify_timm_predict(
		get_testfile('classify_timm', 'img1.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls'
		) == pytest.approx({
		'jia_redian_ruzi_ruzi': 0.9890832304954529, 'siya_ho': 0.005189628805965185,
		'bai_qi-qsr': 0.0015026535838842392, 'kkuem': 0.0012714712647721171,
		'teddy_(khanshin)': 0.00035598213435150683
		}, abs=1e-2)

		def test_classify_timm_predict_2(self):
		assert classify_timm_predict(
		get_testfile('classify_timm', 'img2.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls'
		) == pytest.approx({
		'monori_rogue': 0.6921895742416382, 'stanley_lau': 0.2040979117155075, 'neoartcore': 0.03475344926118851,
		'ayya_sap': 0.005350438412278891, 'goomrrat': 0.004616163671016693
		}, abs=1e-2)

		def test_classify_timm_predict_3(self):
		assert classify_timm_predict(
		get_testfile('classify_timm', 'img3.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls'
		) == pytest.approx({
		'shexyo': 0.9998241066932678, 'oroborus': 0.0001537767384434119, 'jeneral': 7.268482477229554e-06,
		'free_style_(yohan1754)': 3.4537688406999223e-06, 'kakeku': 2.5340586944366805e-06
		}, abs=1e-2)

		def test_classify_timm_predict_4(self):
		assert classify_timm_predict(
		get_testfile('classify_timm', 'img4.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls'
		) == pytest.approx({
		'z.taiga': 0.9999995231628418, 'tina_(tinafya)': 1.2290533391023928e-07,
		'arind_yudha': 6.17258208990279e-08, 'chixiao': 4.949555076905199e-08,
		'zerotwenty_(020)': 4.218352955831506e-08
		}, abs=1e-2)

		def test_classify_timm_predict_5(self):
		assert classify_timm_predict(
		get_testfile('classify_timm', 'img5.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls'
		) == pytest.approx({
		'spam_(spamham4506)': 0.9999998807907104, 'falken_(yutozin)': 4.501828954062148e-08,
		'yuki_(asayuki101)': 3.285677863118508e-08, 'danbal': 5.452678752959628e-09,
		'buri_(retty9349)': 3.757136379789472e-09
		}, abs=1e-2)

		def test_classify_timm_predict_6(self):
		assert classify_timm_predict(
		get_testfile('classify_timm', 'img6.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls'
		) == pytest.approx({
		'mashuu_(neko_no_oyashiro)': 1.0, 'minaba_hideo': 4.543745646401476e-08, 'simosi': 6.499865978781827e-09,
		'maoh_yueer': 4.302619149854081e-09, '7nite': 3.6548184478846224e-09
		}, abs=1e-2)

		def test_classify_timm_predict_1_val(self):
		assert classify_timm_predict(
		get_testfile('classify_timm', 'img1.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls',
		preprocessor='val'
		) == pytest.approx({
		'jia_redian_ruzi_ruzi': 0.9890832304954529, 'siya_ho': 0.005189628805965185,
		'bai_qi-qsr': 0.0015026535838842392, 'kkuem': 0.0012714712647721171,
		'teddy_(khanshin)': 0.00035598213435150683
		}, abs=1e-2)

		def test_classify_timm_predict_1_top_score(self):
		top5, all_, top1 = classify_timm_predict(
		get_testfile('classify_timm', 'img1.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls',
		fmt=('scores-top5', 'scores', 'scores-top1'),
		)
		assert top5 == pytest.approx({
		'jia_redian_ruzi_ruzi': 0.9890832304954529, 'siya_ho': 0.005189628805965185,
		'bai_qi-qsr': 0.0015026535838842392, 'kkuem': 0.0012714712647721171,
		'teddy_(khanshin)': 0.00035598213435150683
		}, abs=1e-2)
		assert top1 == pytest.approx({
		'jia_redian_ruzi_ruzi': 0.9890832304954529,
		}, abs=1e-2)

		assert len(all_) == 9453
		assert all_['jia_redian_ruzi_ruzi'] == pytest.approx(0.9890832304954529, abs=1e-3)

		def test_classify_timm_predict_1_nps(self):
		embedding, logits, prediction = classify_timm_predict(
		get_testfile('classify_timm', 'img1.jpg'),
		repo_id='animetimm/swinv2_base_window8_256.dbv4a-fullxx-cls',
		fmt=('embedding', 'logits', 'prediction'),
		)
		assert embedding.shape == (1024,)
		assert logits.shape == (9453,)
		assert prediction.shape == (9453,)

test/testfile/classify_timm/img1.jpg

0 → 100755

+108 KiB

Loading image diff...

test/testfile/classify_timm/img2.jpg

0 → 100755

+78.8 KiB

Loading image diff...