def test_run_integer_adm_fextractor_akiyo_multiply_enhn_gain_limit_1d2(
self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = IntegerAdmFeatureExtractor(
[asset],
None,
fifo_mode=False,
result_store=None,
optional_dict={'adm_enhn_gain_limit': 1.2})
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['integer_ADM_feature_adm2_score'],
1.116609,
places=6) # float 1.116595
def test_run_integer_adm_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerAdmFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['integer_ADM_feature_adm2_score'],
0.934578625,
places=4)
self.assertAlmostEqual(results[1]['integer_ADM_feature_adm2_score'],
1.000002,
places=6)
def test_run_float_vif_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = FloatVifFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
0.3634208125,
places=6)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
0.7666474166666667,
places=6)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
0.8628533333333334,
places=6)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
0.9159719583333334,
places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale0_score'], 1.0, places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale1_score'], 1.0, places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale2_score'], 1.0, places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale3_score'], 1.0, places=5)
def test_run_integer_psnr_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerPsnrFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['integer_PSNR_feature_psnr_y_score'],
30.755063979166664,
places=4)
self.assertAlmostEqual(
results[0]['integer_PSNR_feature_psnr_cb_score'],
38.4494410625,
places=4)
self.assertAlmostEqual(
results[0]['integer_PSNR_feature_psnr_cr_score'],
40.99191027083334,
places=4)
self.assertAlmostEqual(results[1]['integer_PSNR_feature_psnr_y_score'],
60.0,
places=4)
self.assertAlmostEqual(
results[1]['integer_PSNR_feature_psnr_cb_score'], 60.0, places=4)
self.assertAlmostEqual(
results[1]['integer_PSNR_feature_psnr_cr_score'], 60.0, places=4)
def test_run_integer_psnr_fextractor_16bit(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_16bit_videos_for_testing(
)
self.fextractor = IntegerPsnrFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['integer_PSNR_feature_psnr_y_score'],
32.579806000000005,
places=4)
self.assertAlmostEqual(
results[0]['integer_PSNR_feature_psnr_cb_score'],
39.04694966666667,
places=4)
self.assertAlmostEqual(
results[0]['integer_PSNR_feature_psnr_cr_score'],
41.288954,
places=4)
self.assertAlmostEqual(results[1]['integer_PSNR_feature_psnr_y_score'],
108.0,
places=4)
self.assertAlmostEqual(
results[1]['integer_PSNR_feature_psnr_cb_score'], 108.0, places=4)
self.assertAlmostEqual(
results[1]['integer_PSNR_feature_psnr_cr_score'], 108.0, places=4)
def test_run_integer_motion_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerMotionFeatureExtractor(
[asset, asset_original], None, fifo_mode=False, result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['integer_motion_feature_motion2_score'],
3.895345229166667,
places=8)
self.assertAlmostEqual(
results[1]['integer_motion_feature_motion2_score'],
3.895345229166667,
places=8)
def test_run_float_adm_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = FloatAdmFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['float_ADM_feature_adm2_score'],
0.9345148541666667,
places=4)
self.assertAlmostEqual(results[1]['float_ADM_feature_adm2_score'],
1.0,
places=6)
def test_run_integer_motion_fextractor_forcing_zero(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerMotionFeatureExtractor(
[asset, asset_original],
None,
fifo_mode=False,
result_store=None,
optional_dict={'motion_force_zero': True})
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['integer_motion_feature_motion2_score'], 0.0, places=8)
self.assertAlmostEqual(
results[1]['integer_motion_feature_motion2_score'], 0.0, places=8)
def test_run_float_vif_fextractor_akiyo_multiply_enhn_gain_limit_1d1(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatVifFeatureExtractor(
[asset],
None,
fifo_mode=False,
result_store=None,
optional_dict={'vif_enhn_gain_limit': 1.1},
)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
1.0298451531242514,
places=3)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
1.046596975760772,
places=3)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
1.0485607628500504,
places=3)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
1.0491232394147363,
places=3)
def test_run_float_vif_fextractor_akiyo_multiply_enhn_gain_limit_1(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatVifFeatureExtractor(
[asset],
None,
fifo_mode=False,
result_store=None,
optional_dict={'vif_enhn_gain_limit': 1.0},
)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
0.983699512450884,
places=4)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
0.9974276726830457,
places=4)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
0.9984692380091739,
places=4)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
0.999146211879154,
places=4)
def test_run_float_vif_fextractor_akiyo_multiply(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatVifFeatureExtractor([asset],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
1.0522544319369052,
places=5)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
1.0705609423182443,
places=5)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
1.0731529493098957,
places=5)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
1.0728060231246508,
places=5)
def test_run_integer_vif_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerVifFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale0_score'],
0.3634233125,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale1_score'],
0.7666542916666667,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale2_score'],
0.862872125,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale3_score'],
0.9159956249999999,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale0_score'],
1.000002,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale1_score'],
1.0000023541666667,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale2_score'],
1.0000022916666667,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale3_score'], 1.0, places=5)
def test_run_integer_vif_fextractor_12bit(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_12bit_videos_for_testing(
)
self.fextractor = IntegerVifFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale0_score'],
0.4326293333333333,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale1_score'],
0.8292446666666665,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale2_score'],
0.9067896666666667,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale3_score'],
0.9460693333333333,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale0_score'],
1.000002,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale1_score'],
1.0000023541666667,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale2_score'],
1.0000022916666667,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale3_score'], 1.0, places=5)
class FeatureExtractorTest(unittest.TestCase):
def tearDown(self):
if hasattr(self, 'fextractor'):
self.fextractor.remove_results()
pass
def test_run_float_motion_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = FloatMotionFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_motion_feature_motion2_score'],
3.8953518541666665,
places=8)
self.assertAlmostEqual(
results[1]['float_motion_feature_motion2_score'],
3.8953518541666665,
places=8)
def test_run_integer_motion_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerMotionFeatureExtractor(
[asset, asset_original], None, fifo_mode=False, result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['integer_motion_feature_motion2_score'],
3.895345229166667,
places=8)
self.assertAlmostEqual(
results[1]['integer_motion_feature_motion2_score'],
3.895345229166667,
places=8)
def test_run_float_vif_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = FloatVifFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
0.3634208125,
places=6)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
0.7666474166666667,
places=6)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
0.8628533333333334,
places=6)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
0.9159719583333334,
places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale0_score'], 1.0, places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale1_score'], 1.0, places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale2_score'], 1.0, places=6)
self.assertAlmostEqual(
results[1]['float_VIF_feature_vif_scale3_score'], 1.0, places=5)
def test_run_integer_vif_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerVifFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale0_score'],
0.3634233125,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale1_score'],
0.7666542916666667,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale2_score'],
0.862872125,
places=6)
self.assertAlmostEqual(
results[0]['integer_VIF_feature_vif_scale3_score'],
0.9159956249999999,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale0_score'],
1.000002,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale1_score'],
1.0000023541666667,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale2_score'],
1.0000022916666667,
places=6)
self.assertAlmostEqual(
results[1]['integer_VIF_feature_vif_scale3_score'], 1.0, places=5)
def test_run_float_adm_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = FloatAdmFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['float_ADM_feature_adm2_score'],
0.9345148541666667,
places=4)
self.assertAlmostEqual(results[1]['float_ADM_feature_adm2_score'],
1.0,
places=6)
def test_run_integer_psnr_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerPsnrFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['integer_PSNR_feature_psnr_y_score'],
30.755063979166664,
places=4)
self.assertAlmostEqual(
results[0]['integer_PSNR_feature_psnr_cb_score'],
38.4494410625,
places=4)
self.assertAlmostEqual(
results[0]['integer_PSNR_feature_psnr_cr_score'],
40.99191027083334,
places=4)
self.assertAlmostEqual(results[1]['integer_PSNR_feature_psnr_y_score'],
60.0,
places=4)
self.assertAlmostEqual(
results[1]['integer_PSNR_feature_psnr_cb_score'], 60.0, places=4)
self.assertAlmostEqual(
results[1]['integer_PSNR_feature_psnr_cr_score'], 60.0, places=4)
def test_run_float_adm_fextractor_akiyo_multiply(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatAdmFeatureExtractor([asset],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['float_ADM_feature_adm2_score'],
1.116686,
places=6)
def test_run_float_adm_fextractor_akiyo_multiply_enhn_gain_limit_1(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatAdmFeatureExtractor(
[asset],
None,
fifo_mode=False,
result_store=None,
optional_dict={'adm_enhn_gain_limit': 1.0})
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['float_ADM_feature_adm2_score'],
0.9574308606115118,
places=6)
def test_run_float_adm_fextractor_akiyo_multiply_enhn_gain_limit_1d2(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatAdmFeatureExtractor(
[asset],
None,
fifo_mode=False,
result_store=None,
optional_dict={'adm_enhn_gain_limit': 1.2})
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['float_ADM_feature_adm2_score'],
1.116595,
places=6)
def test_run_float_vif_fextractor_akiyo_multiply(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatVifFeatureExtractor([asset],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
1.0522544319369052,
places=5)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
1.0705609423182443,
places=5)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
1.0731529493098957,
places=5)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
1.0728060231246508,
places=5)
def test_run_float_vif_fextractor_akiyo_multiply_enhn_gain_limit_1(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatVifFeatureExtractor(
[asset],
None,
fifo_mode=False,
result_store=None,
optional_dict={'vif_enhn_gain_limit': 1.0},
)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
0.983699512450884,
places=4)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
0.9974276726830457,
places=4)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
0.9984692380091739,
places=4)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
0.999146211879154,
places=4)
def test_run_float_vif_fextractor_akiyo_multiply_enhn_gain_limit_1d1(self):
ref_path = VmafConfig.test_resource_path(
"yuv",
"refp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
dis_path = VmafConfig.test_resource_path(
"yuv",
"disp_vmaf_hacking_investigation_0_0_akiyo_cif_notyuv_0to0_identity_vs_akiyo_cif_notyuv_0to0_multiply_q_352x288"
)
asset = Asset(dataset="test",
content_id=0,
asset_id=0,
workdir_root=VmafConfig.workdir_path(),
ref_path=ref_path,
dis_path=dis_path,
asset_dict={
'width': 352,
'height': 288
})
self.fextractor = FloatVifFeatureExtractor(
[asset],
None,
fifo_mode=False,
result_store=None,
optional_dict={'vif_enhn_gain_limit': 1.1},
)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale0_score'],
1.0298451531242514,
places=3)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale1_score'],
1.046596975760772,
places=3)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale2_score'],
1.0485607628500504,
places=3)
self.assertAlmostEqual(
results[0]['float_VIF_feature_vif_scale3_score'],
1.0491232394147363,
places=3)
def test_run_integer_adm_fextractor(self):
ref_path, dis_path, asset, asset_original = set_default_576_324_videos_for_testing(
)
self.fextractor = IntegerAdmFeatureExtractor([asset, asset_original],
None,
fifo_mode=False,
result_store=None)
self.fextractor.run()
results = self.fextractor.results
self.assertAlmostEqual(results[0]['integer_ADM_feature_adm2_score'],
0.934578625,
places=4)
self.assertAlmostEqual(results[1]['integer_ADM_feature_adm2_score'],
1.000002,
places=6)