def FetchPredictionsFromFile(filepath):
"""Fetch PredictionOutputs proto from file.
Args:
filepath: local file to load the PredictionOutputs proto
Returns:
PredictionOutputs
"""
pred_outputs = prediction_output_pb2.PredictionOutputs()
with open(filepath, 'rb') as f:
pred_outputs.ParseFromString(f.read())
return pred_outputs
def setUp(self):
super(PredictionDataServiceTest, self).setUp()
tf_ex = tf.train.Example()
feature = tf_ex.features.feature
feature['eeg_channel/num_samples'].int64_list.value.append(20)
feature['eeg_channel/sampling_frequency_hz'].float_list.value.append(
1.0)
for i in range(20):
feature['test_feature'].float_list.value.append(i * 2)
for i in range(20):
feature[('eeg_channel/EEG '
'test_sub-REF/samples')].float_list.value.append(i + 1)
for i in range(20):
feature[('eeg_channel/EEG '
'test_min-REF/samples')].float_list.value.append(1)
time = timestamp_pb2.Timestamp()
time.seconds = 10
time.nanos = 800000000
feature['start_time'].bytes_list.value.append(time.SerializeToString())
feature['segment/patient_id'].bytes_list.value.append(b'test patient')
waveform_data_source = data_source.TfExampleEegDataSource(
tf_ex, 'test key')
pred_outputs = prediction_output_pb2.PredictionOutputs()
pred_output = pred_outputs.prediction_output.add()
pred_output.chunk_info.chunk_id = 'test chunk'
pred_output.chunk_info.chunk_start_time.seconds = 10
pred_output.chunk_info.chunk_start_time.nanos = 800000000
pred_output.chunk_info.chunk_size_sec = 2
label = pred_output.label.add()
label.name = 'test label'
pred_output_2 = pred_outputs.prediction_output.add()
pred_output_2.chunk_info.chunk_id = 'test chunk 2'
pred_output_2.chunk_info.chunk_start_time.seconds = 12
pred_output_2.chunk_info.chunk_start_time.nanos = 800000000
pred_output_2.chunk_info.chunk_size_sec = 2
label = pred_output_2.label.add()
label.name = 'test label'
self._pred = prediction_data_service.PredictionDataService(
pred_outputs, waveform_data_source, 100)
TEST_TF_EX = tf.train.Example()
feature = TEST_TF_EX.features.feature
feature['eeg_channel/num_samples'].int64_list.value.append(4)
feature['eeg_channel/sampling_frequency_hz'].float_list.value.append(1.0)
feature['test_feature'].float_list.value.extend([i * 2 for i in range(20)])
feature['eeg_channel/EEG test_min-REF/samples'].float_list.value.extend(
[1, 2, 3, 4])
feature['eeg_channel/EEG test_sub-REF/samples'].float_list.value.extend(
[1, 1, 1, 1])
feature['raw_label_events'].bytes_list.value.append(
TEST_ANNOTATION.SerializeToString())
feature['start_time'].bytes_list.value.append(TEST_TIME.SerializeToString())
feature['segment/patient_id'].bytes_list.value.append('test id')
TEST_PRED = prediction_output_pb2.PredictionOutputs()
pred_output = TEST_PRED.prediction_output.add()
pred_output.chunk_info.chunk_id = 'test chunk'
pred_output.chunk_info.chunk_start_time.seconds = 10
pred_output.chunk_info.chunk_start_time.nanos = 800000000
pred_output.chunk_info.chunk_size_sec = 2
label = pred_output.label.add()
label.name = 'test label'
label.predicted_value.score = 3
label.actual_value.score = 1
label.attribution_map.attribution.extend(list(range(1, 9)))
label.attribution_map.width = 4
label.attribution_map.height = 2
label.attribution_map.feature_names.extend([
'eeg_channel/EEG test_sub-REF/samples#eeg_channel/EEG test_min-REF/samples',
'eeg_channel/EEG test_min-REF/samples#eeg_channel/EEG test_sub-REF/samples',
