if __name__ == '__main__':
MEAN_FACTOR_NOISE = 0.
STD_FACTOR_NOISE = 0.
NBR_CHANNELS = 3
NEGATIVE_LABEL_ID = 9
# exp number
exp_number = sys.argv[1]
# paths
original_dataset_path = '../../experiments/815/dataset_sequences.h5'
save_path = os.path.join('../../experiments',exp_number)
# create exp folder
createExpFolderandCodeList(save_path)
# set seeds
np.random.seed(0)
random.seed(0)
# load sequence dataset
f = h5py.File(original_dataset_path, 'r')
sequences_x_train = f['sequences_x_train']
sequences_x_valid = f['sequences_x_valid']
# get meta data
nbr_images_in_seq = sequences_x_train.shape[1]
nbr_seq = sequences_x_train.shape[0]
size_x = sequences_x_train[0].shape[1]
size_y = sequences_x_train[0].shape[2]
max_x = 255
range(START_ID + 21, START_ID + 30),
range(START_ID + 31, START_ID + 40),
range(START_ID + 41, START_ID + 50),
range(START_ID + 51, START_ID + 60),
range(START_ID + 61, START_ID + 70),
range(START_ID + 71, START_ID + 80),
range(START_ID + 81, START_ID + 90),
range(START_ID + 91, START_ID + 100)
]
# paths
path_experiments = '../../experiments'
path_save = os.path.join(path_experiments, EXPERIMENT_ID)
# create exp folder
createExpFolderandCodeList(path_save)
# iterate over metrics
for metric in METRICS:
average_df = pd.DataFrame()
for series in series_exp:
average_dict = {}
for risk_level, exp_id in enumerate(series):
df = pd.read_csv(
os.path.join(path_experiments, str(exp_id), 'metrics.csv'))
average_dict[str(risk_level)] = df[metric][0]
# update Dataframe
average_df = average_df.append(average_dict, ignore_index=True)
# compute average and std of averages
mean = average_df.mean()