def test_quadratic_model():
m = QuadraticModel()
w = SimpleWrapper(m)
h = w.fit({'initial_epoch': 0, 'epochs': 5})
compare_dicts(
h, {
'epochs': [e for e in range(0, 3)],
'acc': [4, 1, 0, 1, 4],
'loss': [-4, -1, 0, -1, -4],
'val_acc': [4.5, 1.5, 0.5, 1.5, 4.5],
'val_loss': [-4.5, -1.5, -0.5, -1.5, -4.5]
})
def test_load(test_kernels, label_type, image):
label_file = get_image_label(image, label_type)
isd_str = ale.loads(label_file, props={'kernels': test_kernels[image]})
isd_obj = json.loads(isd_str)
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, image_dict[image]) == []
def test_newhorizons_load(test_kernels):
label_file = get_image_label("lor_0034974380_0x630_sci_1", "isis")
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
isd_obj = json.loads(isd_str)
compare_dict = get_isd('newhorizons')
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_dict) == []
def test_load(test_kernels, usgscsm_comparison_isd):
label_file = get_image_label('M103595705LE')
usgscsm_isd = ale.load(label_file,
props={'kernels': test_kernels},
formatter='usgscsm')
print(usgscsm_isd)
assert compare_dicts(usgscsm_isd, usgscsm_comparison_isd) == []
def test_load_minirf(test_kernels):
label_file = get_image_label('03821_16N196_S1', 'isis3')
# isd_str = ale.loads(label_file, props={'kernels': test_kernels['03821_16N196_S1']}, formatter='usgscsm', verbose=True)
isd_str = ale.loads(label_file, props={'kernels': test_kernels['03821_16N196_S1']}, verbose=True)
isd_obj = json.loads(isd_str)
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, image_dict['03821_16N196_S1']) == []
def test_fit(keras_model):
good_results = {
'acc': [-76, -96, -28],
'val_loss': [4.75, 5.75, 6.75],
'loss': [4.5, 5.5, 6.5],
'val_acc': [4.25, 5.25, 6.25],
'epochs': [4, 5, 6]
}
m = SimpleWrapper(keras_model)
obs = m.fit(fit_args={'initial_epoch': 4, 'epochs': 7})
print('result:\n{}'.format(obs))
print('should be:\n{}'.format(good_results))
compare_dicts(obs, good_results)
def test_cassini_load(test_kernels, usgscsm_compare_dict):
label_file = get_image_label("N1702360370_1")
usgscsm_isd = ale.load(label_file,
props={'kernels': test_kernels},
formatter='usgscsm')
print(usgscsm_isd)
assert compare_dicts(usgscsm_isd, usgscsm_compare_dict) == []
def test_load(test_kernels, label_type, formatter, image):
label_file = get_image_label(image, label_type)
usgscsm_isd_str = ale.loads(label_file, props={'kernels': test_kernels[image]}, formatter=formatter)
usgscsm_isd_obj = json.loads(usgscsm_isd_str)
print(json.dumps(usgscsm_isd_obj, indent=4))
assert compare_dicts(usgscsm_isd_obj, image_dict[image][formatter]) == []
def test_mex_load(test_kernels, formatter, usgscsm_compare_dict, label):
label_file = get_image_label('h5270_0000_ir2', label)
with patch('ale.drivers.mex_drivers.MexHrscPds3NaifSpiceDriver.binary_ephemeris_times', \
new_callable=PropertyMock) as binary_ephemeris_times, \
patch('ale.drivers.mex_drivers.MexHrscPds3NaifSpiceDriver.binary_exposure_durations', \
new_callable=PropertyMock) as binary_exposure_durations, \
patch('ale.drivers.mex_drivers.MexHrscPds3NaifSpiceDriver.binary_lines', \
new_callable=PropertyMock) as binary_lines, \
patch('ale.drivers.mex_drivers.MexHrscIsisLabelNaifSpiceDriver.ephemeris_time', \
new_callable=PropertyMock) as ephemeris_time, \
patch('ale.drivers.mex_drivers.read_table_data', return_value=12345) as read_table_data, \
patch('ale.drivers.mex_drivers.parse_table', return_value={'EphemerisTime': [255744599.02748165, 255744684.33197814, 255744684.34504557], \
'ExposureTime': [0.012800790786743165, 0.012907449722290038, 0.013227428436279297], \
'LineStart': [1, 6665, 6666]}) as parse_table:
ephemeris_time.return_value = [
255744599.02748, 255744623.61901, 255744635.91477
]
binary_ephemeris_times.return_value = [
255744599.02748165, 255744599.04028246, 255744795.73322123
]
binary_exposure_durations.return_value = [
0.012800790786743165, 0.012800790786743165, 0.013227428436279297
]
binary_lines.return_value = [0.5, 1.5, 15086.5]
usgscsm_isd = ale.load(label_file,
props={'kernels': test_kernels},
formatter=formatter,
verbose='true')
assert compare_dicts(
usgscsm_isd,
usgscsm_compare_dict['h5270_0000_ir2'][formatter]) == []
def test_viking1_load(test_kernels, label_type, formatter, image):
label_file = get_image_label(image, label_type)
isis_isd = ale.load(label_file,
props={'kernels': test_kernels[image]},
formatter=formatter)
print(isis_isd)
assert compare_dicts(isis_isd, image_dict[image][formatter]) == []
def test_mro_load(test_kernels, label_type):
label_file = get_image_label('JNCR_2016240_01M06152_V01', label_type)
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
isd_obj = json.loads(isd_str)
compare_dict = get_isd('juno')
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_dict) == []
def test_cassis_load(test_kernels, isis_compare_dict):
label_file = get_image_label(
"CAS-MCO-2016-11-26T22.32.14.582-RED-01000-B1", "isis")
isis_isd = ale.load(label_file,
props={'kernels': test_kernels},
formatter="isis")
assert compare_dicts(isis_isd, isis_compare_dict) == []
def test_load(test_kernels):
label_file = get_image_label('FC21A0038582_15170161546F6F')
compare_dict = get_isd("dawnfc")
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
isd_obj = json.loads(isd_str)
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_dict) == []
def test_cassis_load(test_kernels):
label_file = get_image_label(
"CAS-MCO-2016-11-26T22.32.14.582-RED-01000-B1", "isis")
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
isd_obj = json.loads(isd_str)
compare_dict = get_isd('cassis')
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_dict) == []
def test_load_pds(test_kernels):
label_file = get_image_label("N1702360370_1")
compare_dict = get_isd("cassiniiss")
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
isd_obj = json.loads(isd_str)
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_dict) == []
def test_mro_load(test_kernels, label_type, formatter, isis_compare_dict):
label_file = get_image_label('JNCR_2016240_01M06152_V01', label_type)
usgscsm_isd_str = ale.loads(label_file, props={'kernels': test_kernels}, formatter=formatter)
usgscsm_isd_obj = json.loads(usgscsm_isd_str)
print(json.dumps(usgscsm_isd_obj, indent=4))
assert compare_dicts(usgscsm_isd_obj, isis_compare_dict) == []
def test_mro_load(test_kernels, label_type, formatter, usgscsm_compare_dict, isis_compare_dict):
label_file = get_image_label('B10_013341_1010_XN_79S172W', label_type)
usgscsm_isd_str = ale.loads(label_file, props={'kernels': test_kernels}, formatter=formatter)
usgscsm_isd_obj = json.loads(usgscsm_isd_str)
print(json.dumps(usgscsm_isd_obj, indent=4))
if formatter=='usgscsm':
# Check to change the line based on ISIS vs PDS3
# This is due to a processing step in mroctx2isis that removes samples
# based on some flags on the label
if label_type == 'isis3':
usgscsm_compare_dict['image_samples'] = 5000
assert compare_dicts(usgscsm_isd_obj, usgscsm_compare_dict) == []
else:
assert compare_dicts(usgscsm_isd_obj, isis_compare_dict) == []
def test_load(test_kernels, label_type, image):
label_file = get_image_label(image, label_type)
print(test_kernels)
usgscsm_isd_str = ale.loads(label_file,
props={'kernels': test_kernels},
verbose=True)
usgscsm_isd_obj = json.loads(usgscsm_isd_str)
print(json.dumps(usgscsm_isd_obj, indent=2))
assert compare_dicts(usgscsm_isd_obj, image_dict[image]) == []
def test_hayabusa_load(test_kernels, label_type):
label_file = get_image_label('hyb2_onc_20151203_084458_w2f_l2a',
label_type)
isd_str = ale.loads(label_file,
props={'kernels': test_kernels},
verbose=False)
isd_obj = json.loads(isd_str)
compare_dict = get_isd('hayabusa2')
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_dict) == []
def test_mro_load(test_kernels, label_type):
label_file = get_image_label('B10_013341_1010_XN_79S172W', label_type)
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
isd_obj = json.loads(isd_str)
compare_isd = get_isd('ctx')
if label_type == 'isis3':
compare_isd['image_samples'] = 5000
assert compare_dicts(isd_obj, compare_isd) == []
def test_galileo_load(test_kernels, label_type, kernel_type):
label_file = get_image_label('E6I0032', label_type)
if label_type == 'isis3' and kernel_type == 'isis':
label_file = get_image('E6I0032')
isd_str = ale.loads(label_file)
compare_isd = get_isd('galileossi_isis')
else:
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
compare_isd = get_isd('galileossi')
isd_obj = json.loads(isd_str)
assert compare_dicts(isd_obj, compare_isd) == []
def test_load(test_kernels, label_type, image, kernel_type):
if (kernel_type == "naif"):
label_file = get_image_label(image, label_type)
isd_str = ale.loads(label_file, props={'kernels': test_kernels})
compare_isd = image_dict[image]
else:
label_file = get_image(image)
isd_str = ale.loads(label_file)
compare_isd = get_isd("messmdis_isis")
isd_obj = json.loads(isd_str)
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_isd) == []
def test_load_isis():
label_file = get_image_label("N1702360370_1", label_type="isis3")
compare_dict = get_isd("cassiniiss_isis")
def read_detatched_table(table_label, cube):
return get_table_data("N1702360370_1", table_label["Name"])
with patch('ale.base.data_isis.read_table_data',
side_effect=read_detatched_table):
isd_str = ale.loads(label_file)
isd_obj = json.loads(isd_str)
print(json.dumps(isd_obj, indent=2))
assert compare_dicts(isd_obj, compare_dict) == []
def test_viking1_load(test_kernels, label_type, formatter, image, kernel_type):
if kernel_type == "naif":
label_file = get_image_label(image, label_type)
isis_isd = ale.loads(label_file,
props={'kernels': test_kernels[image]},
formatter=formatter)
compare_dict = image_dict[image][formatter]
else:
label_file = get_image(image)
isis_isd = ale.loads(label_file, verbose=True)
isd_name = image + '_isis'
compare_dict = get_isd(isd_name)
print(isis_isd)
assert compare_dicts(json.loads(isis_isd), compare_dict) == []
def test_linear_state_based_metrics():
m = LinearModel()
w = SimpleWrapper(m)
assert w != m
# History should reflect the change in metrics over for a window
# specified by the epoch range but continuing from where the model
# state was as of the last call to fit().
assert w.eval({}) == (0., 0.)
h = w.fit({'initial_epoch': 0, 'epochs': 3})
compare_dicts(
h, {
'epochs': [e for e in range(0, 3)],
'acc': [1, 2, 3],
'loss': [-1, -2, -3],
'val_acc': [1.5, 2.5, 3.5],
'val_loss': [-1.5, -2.5, -3.5]
})
# Run for a few more epochs
h = w.fit({'initial_epoch': 3, 'epochs': 5})
compare_dicts(
h, {
'epochs': [e for e in range(3, 5)],
'acc': [4, 5],
'loss': [-4, -5],
'val_acc': [4.5, 5.5],
'val_loss': [-4.5, -5.5]
})
# And now run for a few more, but reset the epoch range -- the key here
# is that the metrics should continue relative to the previous state,
# not the epoch range
h = w.fit({'initial_epoch': 1, 'epochs': 3})
compare_dicts(
h, {
'epochs': [e for e in range(1, 4)],
'acc': [6, 7, 8],
'loss': [-6, -7, -8],
'val_acc': [6.5, 7.5, 8.5],
'val_loss': [-6.5, -7.5, -8.5]
})
def test_dawn_load(test_kernels, usgscsm_compare_dict):
label_file = get_image_label('FC21A0038582_15170161546F6F')
usgscsm_isd = ale.load(label_file,
props={'kernels': test_kernels},
formatter='usgscsm')
assert compare_dicts(usgscsm_isd, usgscsm_compare_dict) == []
def test_load_isis(test_kernels, label_type, formatter, image):
label_file = get_image_label(image, label_type)
isis_isd = ale.loads(label_file, props={'kernels': test_kernels[image]}, formatter=formatter, verbose=True)
isis_isd_obj = json.loads(isis_isd)
print(json.dumps(isis_isd_obj, indent=4))
assert compare_dicts(isis_isd_obj, image_dict[image][formatter]) == []
def test_load(test_kernels):
isd = {
'radii': {
'semimajor': 1737.4,
'semiminor': 1737.4,
'unit': 'km'
},
'sensor_position': {
'positions':
np.array([[195493.28614388, 211984.05137213, -1766966.32185819],
[194937.01104181, 211343.39987087, -1767100.47633473],
[194380.68920347, 210702.70097122, -1767234.22276254],
[193824.32093681, 210061.95495794, -1767367.56104995],
[193267.90484827, 209421.16305764, -1767500.49120063],
[192711.44397634, 208780.32318594, -1767633.01318365]]),
'velocities':
np.array([[-1069.71225785, -1231.97438089, -258.37880523],
[-1069.80205939, -1232.06556741, -257.5939851],
[-1069.89161639, -1232.15647191, -256.80910187],
[-1069.98092881, -1232.24709434, -256.02415587],
[-1070.06999666, -1232.33743471, -255.2391471],
[-1070.15881991, -1232.42749298, -254.4540759]]),
'unit':
'm'
},
'sun_position': {
'positions':
np.array([[9.50465237e+10, 1.15903815e+11, 3.78729685e+09]]),
'velocities':
np.array([[285707.13474515, -232731.15884149, 592.91742112]]),
'unit':
'm'
},
'sensor_orientation': {
'quaternions':
np.array([[-0.19095485, -0.08452708, 0.88748467, -0.41080698],
[-0.19073945, -0.08442789, 0.88753312, -0.41082276],
[-0.19052404, -0.08432871, 0.88758153, -0.41083852],
[-0.19030862, -0.08422952, 0.88762988, -0.41085426],
[-0.19009352, -0.08412972, 0.88767854, -0.41086914],
[-0.18987892, -0.08402899, 0.88772773, -0.41088271]])
},
'detector_sample_summing': 1,
'detector_line_summing': 1,
'focal_length_model': {
'focal_length': 72.45
},
'detector_center': {
'line': 0.5,
'sample': 2048.0
},
'starting_detector_line': 0,
'starting_detector_sample': 1,
'focal2pixel_lines': [0, -142.85714285714286, 0],
'focal2pixel_samples': [0, 0, -142.85714285714286],
'optical_distortion': {
'kaguyatc': {
'x':
[-0.0009649900000000001, 0.00098441, 8.5773e-06, -3.7438e-06],
'y':
[-0.0013796, 1.3502e-05, 2.7251e-06, -6.193800000000001e-06]
}
},
'image_lines': 400,
'image_samples': 3208,
'name_platform': 'SELENE-M',
'name_sensor': 'Terrain Camera 1',
'reference_height': {
'maxheight': 1000,
'minheight': -1000,
'unit': 'm'
},
'name_model': 'USGS_ASTRO_LINE_SCANNER_SENSOR_MODEL',
'interpolation_method': 'lagrange',
'line_scan_rate': [[0.5, -1.300000011920929, 0.006500000000000001]],
'starting_ephemeris_time': 292234259.82293594,
'center_ephemeris_time': 292234261.12293595,
't0_ephemeris': -1.300000011920929,
'dt_ephemeris': 0.5200000047683716,
't0_quaternion': -1.300000011920929,
'dt_quaternion': 0.5200000047683716
}
label_file = get_image_label('TC1S2B0_01_06691S820E0465')
with patch('ale.drivers.selene_drivers.KaguyaTcPds3NaifSpiceDriver.reference_frame', \
new_callable=PropertyMock) as mock_reference_frame:
mock_reference_frame.return_value = 'IAU_MOON'
usgscsm_isd = ale.load(label_file,
props={'kernels': test_kernels},
formatter='usgscsm')
print(usgscsm_isd)
assert compare_dicts(usgscsm_isd, isd) == []
def test_load(test_kernels):
isd = {
'radii': {
'semimajor': 1737.4,
'semiminor': 1737.4,
'unit': 'km'
},
'sensor_position': {
'positions':
np.array([[195490.61933009, 211972.79163854, -1766965.21759375],
[194934.34274256, 211332.14192709, -1767099.36601459],
[194378.02078141, 210691.44358962, -1767233.10640484],
[193821.65246371, 210050.69819792, -1767366.43865632],
[193265.23762617, 209409.90567475, -1767499.36279925],
[192708.77531467, 208769.0676026, -1767631.87872367]]),
'velocities':
np.array([[-1069.71186948, -1231.97377674, -258.3672068],
[-1069.80166655, -1232.06495814, -257.58238805],
[-1069.89121909, -1232.15585752, -256.7975062],
[-1069.98052705, -1232.24647484, -256.01256158],
[-1070.06959044, -1232.3368101, -255.2275542],
[-1070.15840923, -1232.42686325, -254.44248439]]),
'unit':
'm'
},
'sun_position': {
'positions':
np.array([[9.50465237e+10, 1.15903815e+11, 3.78729685e+09]]),
'velocities':
np.array([[285707.13474515, -232731.15884149, 592.91742112]]),
'unit':
'm'
},
'sensor_orientation': {
'quaternions':
np.array([[-0.19095485, -0.08452708, 0.88748467, -0.41080698],
[-0.19073945, -0.08442789, 0.88753312, -0.41082276],
[-0.19052404, -0.08432871, 0.88758153, -0.41083852],
[-0.19030862, -0.08422952, 0.88762988, -0.41085426],
[-0.19009352, -0.08412972, 0.88767854, -0.41086914],
[-0.18987892, -0.08402899, 0.88772773, -0.41088271]])
},
'detector_sample_summing': 1,
'detector_line_summing': 1,
'focal_length_model': {
'focal_length': 72.45
},
'detector_center': {
'line': 0.5,
'sample': 2048.0
},
'starting_detector_line': 0,
'starting_detector_sample': 1,
'focal2pixel_lines': [0, -142.85714285714286, 0],
'focal2pixel_samples': [0, 0, -142.85714285714286],
'optical_distortion': {
'kaguyatc': {
'x':
[-0.0009649900000000001, 0.00098441, 8.5773e-06, -3.7438e-06],
'y':
[-0.0013796, 1.3502e-05, 2.7251e-06, -6.193800000000001e-06]
}
},
'image_lines': 400,
'image_samples': 3208,
'name_platform': 'SELENE-M',
'name_sensor': 'Terrain Camera 1',
'reference_height': {
'maxheight': 1000,
'minheight': -1000,
'unit': 'm'
},
'name_model': 'USGS_ASTRO_LINE_SCANNER_SENSOR_MODEL',
'interpolation_method': 'lagrange',
'line_scan_rate': [[0.5, -1.300000011920929, 0.006500000000000001]],
'starting_ephemeris_time': 292234259.82293594,
'center_ephemeris_time': 292234261.12293595,
't0_ephemeris': -1.300000011920929,
'dt_ephemeris': 0.5200000047683716,
't0_quaternion': -1.300000011920929,
'dt_quaternion': 0.5200000047683716
}
label_file = get_image_label('TC1S2B0_01_06691S820E0465')
with patch('ale.drivers.selene_drivers.KaguyaTcPds3NaifSpiceDriver.reference_frame', \
new_callable=PropertyMock) as mock_reference_frame:
mock_reference_frame.return_value = 'IAU_MOON'
usgscsm_isd = ale.load(label_file,
props={'kernels': test_kernels},
formatter='usgscsm')
print(usgscsm_isd)
assert compare_dicts(usgscsm_isd, isd) == []
def test_linear_copy():
from copy import copy
m = LinearModel()
w = SimpleWrapper(m)
w_copy = copy(w)
assert w != w_copy
assert w._model != w_copy._model
assert w.eval({}) == w_copy.eval({})
# Move original model ahead 3 epochs
h = w.fit({'initial_epoch': 0, 'epochs': 3})
compare_dicts(
h, {
'epochs': [e for e in range(0, 3)],
'acc': [1, 2, 3],
'loss': [-1, -2, -3],
'val_acc': [1.5, 2.5, 3.5],
'val_loss': [-1.5, -2.5, -3.5]
})
assert w.eval({}) == (-3, 3)
assert w.eval({}) != w_copy.eval({})
# Move copy ahead 2 epochs
h = w_copy.fit({'initial_epoch': 1, 'epochs': 3})
compare_dicts(
h, {
'epochs': [e for e in range(1, 3)],
'acc': [1, 2],
'loss': [-1, -2],
'val_acc': [1.5, 2.5],
'val_loss': [-1.5, -2.5]
})
assert w_copy.eval({}) == (-2, 2)
assert w.eval({}) != w_copy.eval({})
# Move original ahead 1 epochs
h = w.fit({'initial_epoch': 3, 'epochs': 4})
compare_dicts(
h, {
'epochs': [e for e in range(3, 4)],
'acc': [4],
'loss': [-4],
'val_acc': [4.5],
'val_loss': [-4.5]
})
assert w.eval({}) == (-4, 4)
assert w.eval({}) != w_copy.eval({})
# Move copy ahead 2 epochs
h = w_copy.fit({'initial_epoch': 3, 'epochs': 5})
compare_dicts(
h, {
'epochs': [e for e in range(3, 5)],
'acc': [3, 4],
'loss': [-3, -4],
'val_acc': [3.5, 4.5],
'val_loss': [-3.5, -4.5]
})
assert w_copy.eval({}) == (-4, 4)
assert w.eval({}) == w_copy.eval({})
