-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathtrain_features_segmenter.py
277 lines (209 loc) · 11.9 KB
/
train_features_segmenter.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
# Train ESH model on the Calgary Campinas
import numpy as np
import wandb
import torch
import copy
import torchvision.utils as vutils
from torch.utils.data import DataLoader
from torch import optim
from save_model import save_model
import time
from utils.logger import save_config
from dpipe.torch.functional import weighted_cross_entropy_with_logits
from dpipe.train.policy import Schedule
import torch.nn.functional as F
from models import FeaturesSegmenter
from models import UNet2D
from evaluate import dice_score
from save_model import load_model
from evaluate import sdice
from utils.utils import log_images
from utils.logger import save_config
from datetime import datetime
from IPython import embed
def early_feature_segmentor(dataset_train, dataset_train_dice, dataset_val, config, suffix, wandb_mode, level=3, device=torch.device("cuda:0")):
num_epochs = config.num_epochs
batch_size = config.batch_size
folder_time = datetime.now().strftime("%Y-%m-%d_%I-%M-%S_%p")
n_channels_out = config.n_channels_out
initial_lr = config.lr
wandb_run = wandb.init( project='domain_adaptation', entity='sidra', name = config['model_net_name'] + "_" + suffix +"_"+ folder_time, mode = wandb_mode)
train_loader = DataLoader(dataset_train, batch_size=batch_size,
shuffle=True, num_workers=0, drop_last=True)
best_acc = 0
print("level", level)
# ESH
in_channels = 16 * (2 ** level)
model = FeaturesSegmenter(in_channels=in_channels, out_channels=n_channels_out)
model.cuda(device)
# Base Model
unet_model = UNet2D(n_chans_in=1, n_chans_out=n_channels_out, n_filters_init=16)
unet_model = load_model(config,unet_model )
unet_model.cuda(device)
unet_model.eval()
optimizer = optim.Adam(model.parameters(), lr=initial_lr, weight_decay=0)
save_config(config, suffix,folder_time)
print('----------------------------------------------------------------------')
print(' Training started')
print('----------------------------------------------------------------------')
for epoch in range(1, num_epochs + 1):
model.train()
train_loss_total = 0.0
num_steps = 0
for i, batch in enumerate(train_loader):
input_samples, gt_samples, _ = batch
var_input = input_samples.cuda(device)
var_gt = gt_samples.cuda(device, non_blocking=True)
# level specifies the ESH position
if level == 0:
layer_activations = unet_model.init_path(var_input)
preds = model(layer_activations)
elif level == 1: # level = 1
layer_activations_0 = unet_model.init_path(var_input)
layer_activations_1 = unet_model.down1(layer_activations_0)
logits_ = model(layer_activations_1)
preds = F.interpolate(logits_, scale_factor=2, mode='bilinear')
elif level == 2: # level = 1
layer_activations_0 = unet_model.init_path(var_input)
layer_activations_1 = unet_model.down1(layer_activations_0)
layer_activations_2 = unet_model.down2(layer_activations_1)
logits_ = model(layer_activations_2)
preds = F.interpolate(logits_, scale_factor=4, mode='bilinear')
elif level == 3: # level = 1
#down3
layer_activations_0 = unet_model.init_path(var_input)
layer_activations_1 = unet_model.down1(layer_activations_0)
layer_activations_2 = unet_model.down2(layer_activations_1)
layer_activations_3 = unet_model.down3(layer_activations_2)
logits_ = model(layer_activations_3)
preds = F.interpolate(logits_, scale_factor=8, mode='bilinear')
loss = weighted_cross_entropy_with_logits(preds, var_gt )
train_loss_total += loss.item()
optimizer.zero_grad()
loss.backward()
optimizer.step()
num_steps += 1
if epoch % 10 == 0 and wandb_mode == "online":
#logging images
mask = torch.zeros(size=preds.shape)
mask[preds > 0.5] = 1
log_images(input_samples, mask, gt_samples, epoch, "Train")
train_loss_total_avg = train_loss_total / num_steps
num_steps = 0
print('----------------------------------------------------------------------')
print(' Train Dice Calculation')
print('----------------------------------------------------------------------')
with torch.no_grad():
model.eval()
avg_train_dice = []
for img in range(len(dataset_train_dice)): # looping over all 3D files
train_samples, gt_samples, voxel = dataset_train_dice[img] # Get the ith image, label, and voxel
slices = []
for slice_id, img_slice in enumerate(train_samples): # looping over single img
img_slice = img_slice.unsqueeze(0)
img_slice = img_slice.to(device)
if level == 0:
layer_activations = unet_model.init_path(img_slice)
preds = model(layer_activations)
elif level == 1: # level = 1
#down1
layer_activations_0 = unet_model.init_path(img_slice)
layer_activations_1 = unet_model.down1(layer_activations_0)
logits_ = model(layer_activations_1)
preds = F.interpolate(logits_, scale_factor=2, mode='bilinear')
elif level == 2: # level = 1
#down2
layer_activations_0 = unet_model.init_path(img_slice)
layer_activations_1 = unet_model.down1(layer_activations_0)
layer_activations_2 = unet_model.down2(layer_activations_1)
logits_ = model(layer_activations_2)
preds = F.interpolate(logits_, scale_factor=4, mode='bilinear')
elif level == 3: # level = 1
#down3
layer_activations_0 = unet_model.init_path(img_slice)
layer_activations_1 = unet_model.down1(layer_activations_0)
layer_activations_2 = unet_model.down2(layer_activations_1)
layer_activations_3 = unet_model.down3(layer_activations_2)
logits_ = model(layer_activations_3)
preds = F.interpolate(logits_, scale_factor=8, mode='bilinear')
slices.append(preds.squeeze().detach().cpu())
segmented_volume = torch.stack(slices, dim=0)
slices.clear()
train_dice = sdice(gt_samples.squeeze().numpy()>0,
torch.sigmoid(segmented_volume).numpy() >0.5,
voxel[img])
avg_train_dice.append(train_dice)
if epoch % 10 == 0 and wandb_mode == "online":
# logging images
mask = torch.zeros(size=segmented_volume.shape)
mask[segmented_volume > 0.5] = 1
log_images(train_samples, mask.unsqueeze(1), gt_samples, epoch , "Train_dice")
avg_train_dice = np.mean(avg_train_dice)
print('----------------------------------------------------------------------')
print(' Val Dice Calculation')
print('----------------------------------------------------------------------')
with torch.no_grad():
model.eval()
avg_val_dice = []
total_loss = 0.0
val_loss_total_avg = 0.0
slices = []
for img in range(len(dataset_val)):
input_samples, gt_samples, voxel = dataset_val[img] # Get the ith image, label, and voxel
slices = []
for slice_id, img_slice in enumerate(input_samples):
img_slice = img_slice.unsqueeze(0)
img_slice = img_slice.to(device)
if level == 0:
layer_activations = unet_model.init_path(img_slice)
preds = model(layer_activations)
elif level == 1: # level = 1
# down1
layer_activations_0 = unet_model.init_path(img_slice)
layer_activations_1 = unet_model.down1(layer_activations_0)
logits_ = model(layer_activations_1)
preds = F.interpolate(logits_, scale_factor=2, mode='bilinear')
elif level == 2: # level = 1
# down2
layer_activations_0 = unet_model.init_path(img_slice)
layer_activations_1 = unet_model.down1(layer_activations_0)
layer_activations_2 = unet_model.down2(layer_activations_1)
logits_ = model(layer_activations_2)
preds = F.interpolate(logits_, scale_factor=4, mode='bilinear')
elif level == 3:
# down3
layer_activations_0 = unet_model.init_path(img_slice)
layer_activations_1 = unet_model.down1(layer_activations_0)
layer_activations_2 = unet_model.down2(layer_activations_1)
layer_activations_3 = unet_model.down3(layer_activations_2)
logits_ = model(layer_activations_3)
preds = F.interpolate(logits_, scale_factor=8, mode='bilinear')
slices.append(preds.squeeze().detach().cpu())
val_segmented_volume = torch.stack(slices, dim=0)
slices.clear()
val_loss = weighted_cross_entropy_with_logits(val_segmented_volume.unsqueeze(1), gt_samples)
val_dice = sdice(gt_samples.squeeze().numpy()>0,
torch.sigmoid(val_segmented_volume).numpy() >0.5,
voxel[img])
total_loss += val_loss.item()
avg_val_dice.append(val_dice)
if epoch % 10 == 0 and wandb_mode == "online":
# logging images
mask = torch.zeros(size=val_segmented_volume.shape)
mask[torch.sigmoid(val_segmented_volume) > 0.5] = 1
log_images(input_samples, mask.unsqueeze(1), gt_samples, epoch , "Val_dice", img)
val_loss_total_avg = total_loss / len(dataset_val)
avg_val_dice = np.mean(avg_val_dice)
print(f'Epoch: {epoch}, Train Loss: {train_loss_total_avg}, Train DC: {avg_train_dice}, Valid Loss, {val_loss_total_avg}, Valid DC: {avg_val_dice}')# , train_ind_dice: {train_ind_dice}, valid_ind_dice : {val_ind_dice}')
if avg_val_dice > best_acc:
best_acc = avg_val_dice
print("best_acc- after updation", best_acc)
save_model(model, config, suffix, folder_time)
wandb_run.log({
"Epoch": epoch,
"Train Loss": train_loss_total_avg,
"Train DC": avg_train_dice,
"Valid Loss": val_loss_total_avg,
"Valid DC": avg_val_dice,
})
return model