-
Notifications
You must be signed in to change notification settings - Fork 0
/
damas_fista_main.py
362 lines (287 loc) · 13.5 KB
/
damas_fista_main.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
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
import argparse
import sys
from torch.serialization import validate_cuda_device
import wandb
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.backends.cudnn as cudnn
import numpy as np
import os
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ['CUDA_VISIBLE_DEVICES'] = '3' # 使用 GPU 3
import time
from datasets.dataset import SoundDataset
from utils.utils import AverageMeter, save_model, pyContourf_two, Logger
from networks.damas_fista_net import DAMAS_FISTANet
from loss_function.losses import WingLoss
device_ids = [i for i in range(torch.cuda.device_count())]
def parse_option():
parser = argparse.ArgumentParser(description='DAMAS_FISTANet for sound source in pytorch')
parser.add_argument('--print_freq',
type=int,
default=1,
help='print frequency')
parser.add_argument('--save_freq',
type=int,
default=10,
help='save frequency')
parser.add_argument('--train_dir',
help='The directory used to train the models',
default='./data/One_train.txt', type=str)
parser.add_argument('--test_dir',
help='The directory used to evaluate the models',
default='./data/One_val.txt', type=str)
parser.add_argument('--label_dir',
help='The directory used to evaluate the models',
default='./data/sound_source_data_fixed_distance/NewLabel/', type=str)
parser.add_argument('--save_folder', dest='save_folder',
help='The directory used to save the models',
default='./models/',
type=str)
parser.add_argument('--results_dir',
help='The directory used to save the save image',
default='./img_results/', type=str)
parser.add_argument('--start_epoch', default=1, type=int, metavar='N',
help='manual epoch number (useful on restarts)')
parser.add_argument('--val_epochs', default=9, type=int, metavar='N',
help='number of val epochs to run')
parser.add_argument('--epochs', default=100, type=int, metavar='N',
help='number of total epochs to run')
parser.add_argument('--batch_size', default=256, type=int, help='Batch size for dataloader')
parser.add_argument('--learning_rate', '--learning-rate', default=0.1, type=float,
metavar='LR', help='initial learning rate')
parser.add_argument('--MultiStepLR',
action='store_true',
help='using MultiStepLR')
parser.add_argument('--lr_decay_epochs',
type=str,
default='2, 4',
help='where to decay lr, can be a list')
parser.add_argument('--lr_decay_rate',
type=float,
default=0.01,
help='decay rate for learning rate')
parser.add_argument('--momentum', default=0.9, type=float, metavar='M', help='momentum')
parser.add_argument('--weight_decay', '--wd', default=1e-2, type=float,
metavar='W', help='weight decay (default: 1e-4)')
parser.add_argument('--loss', type=str, default=None, choices=['Wing_loss', 'cross_entropy_loss', 'mse_loss'], help='loss')
parser.add_argument('--LayNo',
default=5,
type=int,
help='iteration nums')
args = parser.parse_args()
iterations = args.lr_decay_epochs.split(',')
args.lr_decay_epochs = list([])
for it in iterations:
args.lr_decay_epochs.append(int(it))
record_time = time.localtime(time.time())
args.model_name = 'DAMAS_FISTA-Net_lr_{}_decay_{}_bsz_{}_{}'.\
format(args.learning_rate, args.weight_decay, args.batch_size, time.strftime('%m-%d-%H-%M', record_time))
if args.MultiStepLR:
args.model_name = '{}_MultiStepLR'.format(args.model_name)
return record_time, args
# 加载声源数据
def set_loader(args):
train_dataloader = torch.utils.data.DataLoader(
SoundDataset(args.train_dir, args.label_dir),
batch_size=args.batch_size, shuffle=True,
num_workers=0, pin_memory=True)
test_dataloader = torch.utils.data.DataLoader(
SoundDataset(args.test_dir, args.label_dir),
batch_size=1, shuffle=True,
num_workers=0, pin_memory=True)
return train_dataloader, test_dataloader
def set_model(args):
# 加载模型
model = DAMAS_FISTANet(args.LayNo)
# 定义loss函数
# criterion = torch.nn.L1Loss(reduction="mean") * 1e6
if args.loss == 'Wing_loss':
criterion = WingLoss()
criterion.cuda()
elif args.loss == 'cross_entropy_loss':
criterion = nn.CrossEntropyLoss()
criterion.cuda()
elif args.loss == 'mse_loss':
criterion = torch.nn.MSELoss()
wandb.watch(model)
else:
criterion = None
model.cuda()
cudnn.benchmark = True
return model, criterion
def set_optimizer(args, model):
# 定义优化器
# optimizer = torch.optim.SGD(model.parameters(), args.learning_rate,
# momentum=args.momentum,
# weight_decay=args.weight_decay)
optimizer = torch.optim.Adam(model.parameters(), lr=args.learning_rate, betas=(0.9, 0.999), eps=1e-8, weight_decay=args.weight_decay)
return optimizer
def adjust_learning_rate(args, optimizer, epoch):
# 定义学习率策略
if args.MultiStepLR:
# args.learning_rate *= 0.95 ** (epoch)
args.learning_rate *= 0.95
for param_group in optimizer.param_groups:
param_group['lr'] = args.learning_rate
print('lr=', param_group['lr'])
def train(train_dataloader, model, criterion, optimizer, epoch, args):
model.train()
batch_time = AverageMeter()
data_time = AverageMeter()
losses = AverageMeter()
end = time.time()
for idx, (ATA, ATb, DAS_results, label, label_coordinate_to_vector_pos, _) in enumerate(train_dataloader):
# x0 = torch.zeros(DAS_results.shape, dtype=torch.float64)
data_time.update(time.time() - end)
if torch.cuda.is_available():
ATA = ATA.cuda(non_blocking=True)
ATb = ATb.cuda(non_blocking=True)
label = label.cuda(non_blocking=True)
# x0 = x0.cuda(non_blocking=True)
DAS_results = DAS_results.cuda(non_blocking=True)
bsz = label.shape[0]
# compute_loss
# output = model(x0, ATA, ATb)
output = model(DAS_results, ATA, ATb)
print("==================output", torch.max(output))
# print("==================label", torch.max(label))
# print("==================torch.sqrt(output)", torch.sqrt(output))
# SPL_output = 20 * torch.log10(2.2204e-16 + torch.sqrt(output) / 2e-5)
# SPL_output = torch.clamp(SPL_output, min=0.0)
# print("==================output", output)
# SPL_label = 20 * torch.log10(2.2204e-16 + torch.sqrt(label) / 2e-5)
# SPL_label = torch.clamp(SPL_label, min=0.0)
# print("==================label", label)
# SPL_label.shape_____[batch, 1681, 1]
# center_points = torch.zeros(SPL_output.shape[0], 1)
# for i in range(output.shape[0]):
# center_points[i] = (torch.squeeze(SPL_output[i], 1)[torch.squeeze(label_coordinate_to_vector_pos[i], 1).type(torch.long)]
# - torch.squeeze(SPL_label[i], 1)[torch.squeeze(label_coordinate_to_vector_pos[i], 1).type(torch.long)]) ** 2
loss = torch.sum((output - label) ** 2)
# loss = torch.mean(center_points)
# loss = 0.5 * torch.sum((SPL_output - SPL_label) ** 2) / torch.sum(SPL_label ** 2)
# loss = 0.5 * torch.sum((output - label) ** 2) / torch.sum(label ** 2)
# loss = criterion(SPL_output, SPL_label)
# loss = torch.sum((SPL_output - SPL_label) ** 2)
# loss = torch.mean((output - label) ** 2)
loss = 1 * torch.sum((output - label) ** 2)
# loss = torch.abs(torch.mean((output-label))) / 2e-5 / 2e-5
losses.update(loss.item(), bsz)
# compute gradient and do Adam step
optimizer.zero_grad()
loss.backward()
# torch.nn.utils.clip_grad_norm_(model.parameters(), 5.0)
optimizer.step()
# print("==================================grad")
# print(model.lambda_step.grad)
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
# print info
if (idx + 1) % args.print_freq == 0:
print('Train: [{0}][{1}/{2}]\t'
'BT {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'DT {data_time.val:.3f} ({data_time.avg:.3f})\t'
'loss {loss.val:.3f} ({loss.avg:.3f})'.format(
epoch,
idx + 1,
len(train_dataloader),
batch_time=batch_time,
data_time=data_time,
loss=losses))
sys.stdout.flush()
return losses.avg
def test(test_dataloader, model, criterion, args, record_time, epoch):
"""test"""
# args.results_dir = args.results_dir + '{}__ckpt_epoch_{}/'.format(time.strftime('%m-%d-%H-%M', record_time), epoch)
# if not os.path.exists(args.results_dir):
# os.makedirs(args.results_dir)
# filename = args.results_dir + 'loss.txt'
model.eval()
batch_time = AverageMeter()
losses = AverageMeter()
with torch.no_grad():
end = time.time()
for idx, (ATA, ATb, DAS_results, label, _, sample_name) in enumerate(test_dataloader):
# x0 = torch.zeros(DAS_results.shape, dtype=torch.float64)
ATA = ATA.cuda()
ATb = ATb.cuda()
DAS_results = DAS_results.cuda()
label = label.cuda()
# x0 = x0.cuda()
bsz = label.shape[0]
# forward
output = model(DAS_results, ATA, ATb)
SPL_output = 20 * torch.log10(2.2204e-16 + torch.sqrt(output) / 2e-5)
SPL_output = torch.clamp(SPL_output, min=0.0)
SPL_label = 20 * torch.log10(2.2204e-16 + torch.sqrt(label) / 2e-5)
SPL_label = torch.clamp(SPL_label, min=0.0)
# loss = criterion(SPL_output, SPL_label)
# loss = torch.mean((SPL_output - SPL_label) ** 2)
loss = torch.sum((output - label) ** 2)
# np_loss = loss.cpu().numpy()
# np_output = output.cpu().numpy()
# np_label = label.cpu().numpy()
print("##MAX_SPL_output", torch.max(SPL_output))
print("####MAX_SPL_label", torch.max(SPL_label))
# update metric
losses.update(loss.item(), bsz)
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if idx % args.print_freq == 0:
print('Test: [{0}/{1}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(
idx,
len(test_dataloader),
batch_time=batch_time,
loss=losses))
return losses.avg
def main():
record_time, args = parse_option()
sys.stdout = Logger("train_info_debug_{}.txt".format(time.strftime('%m-%d-%H-%M', record_time)))
args.save_folder = args.save_folder + '{}/'.format(time.strftime('%m-%d-%H-%M', record_time))
if not os.path.exists(args.save_folder):
os.makedirs(args.save_folder)
# wandb.init(
# project='damas_fista_sound_source_location',
# entity='joaquin_chou',
# name="DAMAS_FISTANet" + args.model_name,
# config=args
# )
# build data loader
train_dataloader, test_dataloader = set_loader(args)
# build model and criterion
model, criterion = set_model(args)
# build optimizer
optimizer = set_optimizer(args, model)
print('===========================================')
print('DAMAS_FISTA-Net...')
print('===> Start Epoch {} End Epoch {}'.format(args.start_epoch, args.epochs + 1))
# training routine
for epoch in range(1, args.epochs + 1):
if args.MultiStepLR:
adjust_learning_rate(args, optimizer, epoch)
# train for one epoch
time1 = time.time()
loss = train(train_dataloader, model, criterion, optimizer,
epoch, args)
time2 = time.time()
print('epoch {}, total time {:.2f}'.format(epoch, time2 - time1))
# wandb.log({"train_loss": loss, "epoch": epoch})
if epoch % args.save_freq == 0:
save_file = os.path.join(
args.save_folder, 'ckpt_epoch_{epoch}.pth'.format(epoch=epoch))
save_model(model, optimizer, args, epoch, save_file)
# evaluation
if epoch % args.val_epochs == 0:
loss = test(test_dataloader, model, criterion, args, record_time, epoch)
# wandb.log({"test_loss": loss, "epoch": epoch})
# save the last model
save_file = os.path.join(args.save_folder, 'last.pth')
save_model(model, optimizer, args, args.epochs, save_file)
if __name__ == '__main__':
main()