-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain.py
186 lines (156 loc) · 7.95 KB
/
train.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
import numpy as np
import os
import torch
import torchvision.models as models
from torchvision import transforms
import pytorch_lightning as pl
import wandb
from pytorch_lightning.loggers import WandbLogger
from random import randrange
from os import listdir
from os.path import isfile, join
from dataset import HeartData
from unet import UVnet
import argparse
def main(params):
# set params
batch_size = params.batch_size
learning_rate = params.learning_rate
epochs = params.epochs
gpu_id = params.gpu_id
seed = params.seed
if params.loss == 0:
loss = "DICE"
elif params.loss == 1:
loss = "BCE"
if params.optimiser == 0:
optimiser = "Adam"
elif params.optimiser == 1:
optimiser = "SGD"
# determinism
# np.random.seed(seed) # numpy
# torch.random.manual_seed(seed)
# torch.cuda.manual_seed(seed)
# torch.backends.cudnn.deterministic = True
# torch.backends.cudnn.benchmark = False
################################################################################################################
# set up logging ###############################################################################################
################################################################################################################
name_base = "full_dataset"
name_lr = "_lr" + str(learning_rate)
name_bs = "_bs" + str(batch_size)
name_e = "_e" + str(epochs)
rand_id = "_" + str(randrange(1111111111, 9999999999))
name = name_base + name_lr + name_bs + name_e + rand_id
wandb_logger = WandbLogger(name=name, project="sweep")
################################################################################################################
# define hyper-parameters ######################################################################################
################################################################################################################
hparams = {
"batch_size": batch_size,
"learning_rate": learning_rate,
"epochs": epochs,
"training": ["ct_2", "ct_3", "ct_4", "ct_5", "ct_9"], # 1 means 10
"validation": ["ct_1", "ct_8"],
"seed": seed,
"loss": loss,
"optimiser": optimiser,
"normalisation": "no",
"random": "yes"
}
################################################################################################################
# set up datasets and loading ##################################################################################
################################################################################################################
cts = {
"ct_2": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_2_la.nii",
"ct_3": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_3_la.nii",
"ct_4": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_4_la.nii",
"ct_5": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_5_la.nii",
"ct_6": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_6_la.nii",
"ct_7": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_7_la.nii",
"ct_8": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_8_la.nii",
"ct_9": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_9_la.nii",
"ct_1": "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/cts/ct_10_la.nii"
}
# load DRR filenames
path_to_files = "/home/fryderyk/Desktop/Datasets/STACOM_SLAWT/STACOM_3d_64/"
train_files = []
val_files = []
for f in listdir(path_to_files): # list all files in directory
if isfile(join(path_to_files, f)):
value = [True for key in hparams["training"] if key in f] # check if a training ct is in the filename
if value: # if value is not empty
train_files.append(f)
value = [True for key in hparams["validation"] if key in f] # check if a validation ct is in the filename
if value:
val_files.append(f)
train_files = sorted(train_files)
val_files = sorted(val_files)
# create train and val dictionaries
train_data = {
"path_files": path_to_files,
"input": train_files,
"target": cts
}
val_data = {
"path_files": path_to_files,
"input": val_files,
"target": cts
}
# create datasets
train_data_set = HeartData(train_data)
val_data_set = HeartData(val_data)
train_dataloader = torch.utils.data.DataLoader(train_data_set, batch_size=hparams["batch_size"], shuffle=True,
num_workers=12)
val_dataloader = torch.utils.data.DataLoader(val_data_set, batch_size=hparams["batch_size"], shuffle=False,
num_workers=12)
################################################################################################################
# set up model####################################################################################################
################################################################################################################
model = UVnet(hparams=hparams)
################################################################################################################
# train the model ##############################################################################################
################################################################################################################
# early stopping
early_stop_callback = pl.callbacks.EarlyStopping(
monitor='val_loss', # what is monitored
min_delta=0.0, # an absolut change smaller than that is not an improvement
patience=4, # number of validation epochs with no improvement
mode='min' # when quantity stops decreasing
)
# set up trainer
trainer = pl.Trainer(
logger=wandb_logger,
max_epochs=hparams["epochs"],
callbacks=[early_stop_callback],
gpus=[gpu_id] if torch.cuda.is_available() else None # write -1 to use all GPUs
)
# train
trainer.fit(model, train_dataloader, val_dataloader)
################################################################################################################
# save model ###################################################################################################
################################################################################################################
save_path = "/home/fryderyk/Desktop/repository/models/" + name + ".pt"
model = model.cpu()
# save entire model
# torch.save(model, save_path)
# save model weights for inference
torch.save(model.state_dict(), save_path)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("-b", "--batch_size", type=int, default=10,
help="Batch size (int). Default: 10")
parser.add_argument("-l", "--learning_rate", type=float, default=0.0001,
help="Learning rate (float). Default: 0.0001")
parser.add_argument("-e", "--epochs", type=int, default=15,
help="Amount of epochs (int)")
parser.add_argument("-g", "--gpu_id", type=int, default=0, choices=[0, 1],
help="GPU ID (int), either 0 or 1. Default: 0")
parser.add_argument("-s", "--seed", type=int, default=1,
help="Setting the random seed. Default: 1")
parser.add_argument("-ls", "--loss", type=int, default=0, choices=[0, 1],
help="Choose loss function. 0:DICE, 1:WeightedPixBCE. Default: 0")
parser.add_argument("-op", "--optimiser", type=int, default=0, choices=[0, 1],
help="Choose optimiser. 0:Adam, 1:SGD. Default: 0")
args = parser.parse_args()
main(args)