import torch
import sys
sys.path.append("..")
# ===================
from OCT_train import trainModels
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
trainModels(model='SOASNet_single',
data_set='duke',
input_dim=1,
epochs=250,
width=64,
depth=4,
depth_limit=6,
repeat=5,
l_r=1e-3,
l_r_s=True,
train_batch=4,
shuffle=True,
loss='ce',
norm='bn',
log='MICCAI_Duke_Results',
class_no=8,
cluster=True,
data_augmentation_train='all',
data_augmentation_test='none')
import torch
import sys
sys.path.append("..")
# ===================
from OCT_train import trainModels
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
if __name__ == '__main__':
#
trainModels(model='SOASNet',
data_set='ours',
input_dim=1,
epochs=50,
width=16,
depth=4,
depth_limit=6,
repeat=3,
l_r=1e-3,
l_r_s=True,
train_batch=4,
shuffle=True,
loss='ce',
norm='bn',
log='MICCAI_Our_Data_Results',
class_no=2,
cluster=True,
data_augmentation_train='all',
data_augmentation_test='all')
print('Finished.')
# cluster=False)
# #
# ====================================
# SegNet based
# ====================================
#
trainModels(model='SOASNet_very_large_kernel',
data_set='ours',
input_dim=1,
epochs=1,
width=16,
depth=4,
depth_limit=6,
repeat=1,
l_r=1e-3,
l_r_s=True,
train_batch=4,
shuffle=True,
loss='ce',
norm='bn',
log='Test',
class_no=2,
cluster=False,
data_augmentation_train='all',
data_augmentation_test='none')
#
# trainModels(model='RelayNet',
# input_dim=1,
# epochs=250,
# width=64,
# depth=4,
