-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathbase_options.py
208 lines (182 loc) · 11.4 KB
/
base_options.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
import argparse
import os
import sys
import torch
import datasets
from seg_utils import utils
IS_DEBUG = sys.gettrace() is not None
class BaseOptions:
"""This class defines options used during both training and test time.
It also implements several helper functions such as parsing, printing, and saving the options.
It also gathers additional options defined in <modify_commandline_options> functions in both dataset class and models class.
"""
def __init__(self, save=True, display=True):
"""Reset the class; indicates the class hasn't been initailized"""
self.initialized = False
self.save = save
self.display = display
def initialize(self, parser):
"""Define the common options that are used in both training and test."""
# basic parameters
parser.add_argument('--dataroot', required=True,
help='path to images (should have subfolders trainA, trainB, valA, valB, etc)')
parser.add_argument('--name', type=str, default='run_name',
help='name of this run. It decides where to store samples and models')
parser.add_argument('--experiment', type=str, default="other_experiments",
help="experiments are collections of runs. Sets subdir within checkpoints dir and mlflow param ")
parser.add_argument('--gpu_ids', type=str, default='0', help='gpu ids: e.g. 0 0,1,2, 0,2. use -1 for CPU')
parser.add_argument('--checkpoints_dir', type=str, default='./checkpoints', help='models are saved here')
parser.add_argument('--phase', type=str, default='train',
help='train | val | test. Expects a dataset subfolder of the same name.')
# models parameters
parser.add_argument('--model', type=str, default='unet_128',
help='chooses which models to use. [unet_128 | unet_256 | resnet_6blocks | resnet_9blocks]')
parser.add_argument('--input_nc', type=int, default=1,
help='# of input image channels: 3 for RGB and 1 for grayscale')
parser.add_argument('--output_nc', type=int, default=1,
help='# of output image channels: 3 for RGB and 1 for grayscale')
parser.add_argument('--filters', type=int, default=64, help='# of filters in the last conv layer')
parser.add_argument('--norm', type=str, default='instance',
help='instance normalization or batch normalization [instance | batch | none]')
parser.add_argument('--init_type', type=str, default='normal',
help='network initialization [normal | xavier | kaiming | orthogonal]')
parser.add_argument('--init_gain', type=float, default=0.02,
help='scaling factor for normal, xavier and orthogonal.')
parser.add_argument('--no_dropout', action='store_true', help='no dropout for the generator')
# parser.add_argument('--bbox_loss_weighting', default=0, type=float, help="if >0 than add bbox as an objective")
# dataset parameters
parser.add_argument('--dataset_mode', type=str, default='segmentation',
help='chooses how datasets are loaded. [segmentation | infer]')
parser.add_argument('--serial_batches', action='store_true',
help='if true, takes images in order to make batches, otherwise takes them randomly')
parser.add_argument('--num_threads', default=4, type=int, help='# threads for loading data')
parser.add_argument('--batch_size', type=int, default=2, help='input batch size')
parser.add_argument('--load_size', type=int, default=286, help='scale images to this size')
parser.add_argument('--crop_size', type=int, default=256, help='then crop to this size')
parser.add_argument('--max_dataset_size', type=int, default=None,
help='Maximum number of samples allowed per dataset. If the dataset directory contains '
'more than max_dataset_size, only a subset is loaded.')
parser.add_argument('--max_infer_dataset_size', type=int, default=None,
help='Max samples allowed for the inference dataset. Defaults to max_dataset_size.')
parser.add_argument('--preprocess', type=str, default=['resize", "crop'], nargs="*",
help='preprocessing steps. Any combination of: [resize | crop | affine | norm | gamma]'
'separate by space.')
parser.add_argument('--display_winsize', type=int, default=256,
help='display window size for both visdom and HTML')
# additional parameters
parser.add_argument('--verbose', action='store_true', help='if specified, print more debugging information')
parser.add_argument('--suffix', default='', type=str,
help='customized suffix: opt.name = opt.name + suffix: e.g., {models}_{netG}_size{load_size}')
parser.add_argument('--metrics', default="dice", type=str, nargs="*",
help="metrics to use, any combination of: [iou | dice | curvature | shape | hausdorff |"
"simplicity | convexity | cosine | all]. Use all to get everything. separate by space")
parser.add_argument('--loss', default="vanilla", type=str, nargs='*',
help="loss to use (can be multiple separated by space) Options are: "
"dice | tversky | focal | crossentropy | vanilla | bbox. separate by space.")
parser.add_argument('--weights', default=None, type=float, nargs='*',
help="weight for each loss function. Should match length of losses. Default is 1 for all."
"separate by space.")
parser.add_argument("--include_pseudo", default=False, action="store_true",
help="If true than will add the psuedo image as a channel. This means a pseudo image must "
"be generated using the CycleGAN")
parser.add_argument('--coord_conv', default=False, action="store_true", help="include coordinate convolution")
parser.add_argument("--label_vals", type=str, default="all",
help="choose to only use one specific label. If used then output_nc should be set to 2."
"Options are all, lv, lv_epi, or la.")
parser.add_argument("--data_filter", type=str, default=(), nargs="*",
help="only include data files with these strings in the file name")
parser.add_argument('--norm_dataroot', type=str, default=None,
help='path to the dataset used for mean normalization if desired. Else uses the current dataset.')
self.initialized = True
return parser
def gather_options(self, inp_str=None):
"""Initialize our parser with basic options(only once).
Add additional models-specific and dataset-specific options.
These options are defined in the <modify_commandline_options> function
in models and dataset classes.
"""
if not self.initialized: # check if it has been initialized
parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser = self.initialize(parser)
# get the basic options
opt, _ = parser.parse_known_args(inp_str)
# modify models-related parser options
# model_name = opt.model
# model_option_setter = models.get_option_setter(model_name)
# parser = model_option_setter(parser, self.isTrain)
# opt, _ = parser.parse_known_args() # parse again with new defaults
# modify dataset-related parser options
dataset_name = opt.dataset_mode
dataset_option_setter = datasets.get_option_setter(dataset_name)
parser = dataset_option_setter(parser, self.isTrain)
# save and return the parser
self.parser = parser
return self.parser
def print_options(self, opt):
"""Print and save options
It will print both current options and default values(if different).
It will save options into a text file / [checkpoints_dir] / opt.txt
"""
message = ''
message += '----------------- Options ---------------\n'
for k, v in sorted(vars(opt).items()):
comment = ''
default = self.parser.get_default(k)
if v != default:
comment = '\t[default: %s]' % str(default)
message += '{:>25}: {:<30}{}\n'.format(str(k), str(v), comment)
message += '----------------- End -------------------'
print(message)
if self.save:
# save to the disk
expr_dir = os.path.join(opt.checkpoints_dir, opt.experiment, opt.name)
utils.mkdirs(expr_dir)
file_name = os.path.join(expr_dir, '{}_opt.txt'.format(opt.phase))
with open(file_name, 'wt') as opt_file:
opt_file.write(message)
opt_file.write('\n')
def parse(self, inp_str=None):
"""Parse our options, create checkpoints directory suffix, and set up gpu device."""
opt = self.gather_options(inp_str).parse_args(inp_str)
opt.isTrain = self.isTrain # train or test
if IS_DEBUG:
if self.display:
print("Detected DEBUG mode, setting num_threads to 0")
opt.num_threads = 0
opt.include_bbox = "bbox" in opt.loss
opt.include_adversarial = "adversarial" in opt.loss
# process opt.suffix
if opt.suffix:
suffix = ('_' + opt.suffix.format(**vars(opt))) if opt.suffix != '' else ''
opt.name = opt.name + suffix
if opt.label_vals != "all":
if opt.output_nc > 2:
print(f"single label_val specified ({opt.label_vals}), but output_nc is {opt.output_nc}, setting to 2")
opt.output_nc = 2
if not self.isTrain or (self.isTrain and opt.restore_filename is not None):
opt.restore_file = os.path.join(opt.restore_experiment, opt.restore_filename)
else:
opt.restore_file = None
# set gpu ids
str_ids = opt.gpu_ids.split(',')
opt.gpu_ids = []
for str_id in str_ids:
id = int(str_id)
if id >= 0:
opt.gpu_ids.append(id)
if len(opt.gpu_ids) > 0:
torch.cuda.set_device(opt.gpu_ids[0])
# check other options for consistency
if opt.include_pseudo:
pseudo_dir = os.path.join(opt.dataroot, opt.phase, "pseudos")
assert os.path.exists(pseudo_dir), f"--include_pseudo specified but no pseudo dir at {pseudo_dir}"
if opt.include_adversarial:
assert opt.adversarial_dataset_root is not None, "adversarial loss is included but no dataset specified"
if os.path.basename(opt.dataroot) == "":
opt.dataroot = os.path.split(opt.dataroot)[0] # remove trailing slashes
if opt.isTrain and opt.infer_dataset_root is not None and os.path.basename(opt.infer_dataset_root) == "":
opt.infer_dataset_root = os.path.split(opt.infer_dataset_root)[0] # remove trailing slashes
if self.display:
self.print_options(opt)
self.opt = opt
return self.opt