-
Notifications
You must be signed in to change notification settings - Fork 1
/
train_parallel_multi.py
179 lines (168 loc) · 7.37 KB
/
train_parallel_multi.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
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "1,0"
import torch
import torch.nn.functional as F
import sys
import numpy as np
from datetime import datetime
from torchvision.utils import make_grid
from Models.model_multi import TMSOD
from Models.data_multi import SalObjDataset, test_dataset
from Models.utils import clip_gradient, adjust_lr
from tensorboardX import SummaryWriter
import logging
import torch.backends.cudnn as cudnn
from options import opt
import torch.nn as nn
from smooth_loss import get_saliency_smoothness
import argparse
import Models.misc as utils
from Models.default import _C as cfg
import random
from torch.utils.data import DataLoader
import Models.samplers as samplers
cudnn.benchmark = True
cfg = cfg
utils.init_distributed_mode(cfg.TRAIN)
device = torch.device(cfg.TRAIN.device)
seed = cfg.TRAIN.seed + utils.get_rank()
torch.manual_seed(seed)
np.random.seed(seed)
random.seed(seed)
model = TMSOD()
if (opt.load is not None):
model.load_state_dict({k.replace('module.',''):v for k,v in torch.load(opt.load, map_location='cuda:0').items()}, strict=False)
print('load model from ', opt.load)
model.to(device)
params = model.parameters()
optimizer = torch.optim.AdamW(params, lr=opt.lr, weight_decay=cfg.TRAIN.weight_decay)
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer, cfg.TRAIN.lr_drop)
train_image_root = opt.rgb_label_root
train_gt_root = opt.gt_label_root
train_depth_root = opt.depth_label_root
val_image_root = opt.val_rgb_root
val_gt_root = opt.val_gt_root
val_depth_root = opt.val_depth_root
save_path = opt.save_path
if not os.path.exists(save_path):
os.makedirs(save_path)
print('load data...')
dataset_train = SalObjDataset(train_image_root, train_gt_root, train_depth_root, trainsize=opt.trainsize)
dataset_test = test_dataset(val_image_root, val_gt_root, val_depth_root, opt.trainsize)
if cfg.TRAIN.distributed:
sampler_train = samplers.DistributedSampler(dataset_train)
else:
sampler_train = torch.utils.data.RandomSampler(dataset_train)
batch_sampler_train = torch.utils.data.BatchSampler(
sampler_train, opt.batchsize, drop_last=True)
train_loader = DataLoader(dataset_train, batch_sampler=batch_sampler_train,
num_workers=16,
pin_memory=True)
total_step = len(train_loader)
if cfg.TRAIN.distributed:
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[cfg.TRAIN.gpu], find_unused_parameters=True)
logging.basicConfig(filename=save_path + 'log.log', format='[%(asctime)s-%(filename)s-%(levelname)s:%(message)s]',
level=logging.INFO, filemode='a', datefmt='%Y-%m-%d %I:%M:%S %p')
logging.info("BBSNet_unif-Train")
logging.info("Config")
logging.info(
'epoch:{};lr:{};batchsize:{};trainsize:{};clip:{};decay_rate:{};load:{};save_path:{};decay_epoch:{}'.format(
opt.epoch, opt.lr, opt.batchsize, opt.trainsize, opt.clip, opt.decay_rate, opt.load, save_path,
opt.decay_epoch))
CE = torch.nn.BCEWithLogitsLoss()
step = 0
best_mae = 1
best_epoch = 0
print(len(train_loader))
class SoftDiceLoss(nn.Module):
def __init__(self, weight=None, size_average=True):
super(SoftDiceLoss, self).__init__()
def forward(self, logits, targets):
num = targets.size(0)
smooth = 1
probs = torch.sigmoid(logits)
m1 = probs.view(num, -1)
m2 = targets.view(num, -1)
intersection = (m1 * m2)
score = 2. * (intersection.sum(1) + smooth) / (m1.sum(1) + m2.sum(1) + smooth)
score = 1 - score.sum() / num
return score
dice = SoftDiceLoss()
def train(train_loader, model, optimizer, epoch, save_path):
global step
model.train()
loss_all = 0
epoch_step = 0
if cfg.TRAIN.distributed:
sampler_train.set_epoch(epoch)
try:
for i, (images, gts, depths) in enumerate(train_loader, start=1):
optimizer.zero_grad()
images = images.to(device)
depths = depths.to(device)
gts = gts.to(device)
out = model(images, depths)
sml = get_saliency_smoothness(torch.sigmoid(out), gts)
loss1_fusion = F.binary_cross_entropy_with_logits(out, gts)
dice_loss = dice(out, gts)
loss_seg = loss1_fusion + sml + dice_loss
loss = loss_seg
loss.backward()
clip_gradient(optimizer, opt.clip)
optimizer.step()
step += 1
epoch_step += 1
loss_all += loss.data
if i % 50 == 0 or i == total_step or i == 1:
print(
'{} Epoch [{:03d}/{:03d}], Step [{:04d}/{:04d}], loss: {:.4f} sml: {:.4f} loss1_fusion: {:0.4f} dice: {:0.4f}'.
format(datetime.now(), epoch, opt.epoch, i, total_step, loss.data, sml.data, loss1_fusion.data,
dice_loss.data))
logging.info(
'#TRAIN#:Epoch [{:03d}/{:03d}], Step [{:04d}/{:04d}], loss: {:.4f} sml: {:.4f} loss1_fusion: {:0.4f} dice: {:0.4f}'.
format(epoch, opt.epoch, i, total_step, loss.data, sml.data, loss1_fusion.data, dice_loss.data))
loss_all /= epoch_step
logging.info('#TRAIN#:Epoch [{:03d}/{:03d}], Loss_AVG: {:.4f}'.format(epoch, opt.epoch, loss_all))
if (epoch) % 10 == 0 and epoch >= 0:
torch.save(model.module.state_dict(), save_path + 'UniSOD_epoch_{}.pth'.format(epoch))
except KeyboardInterrupt:
print('Keyboard Interrupt: save model and exit.')
if not os.path.exists(save_path):
os.makedirs(save_path)
torch.save(model.module.state_dict(), save_path + 'UniSOD_epoch_{}.pth'.format(epoch + 1))
print('save checkpoints successfully!')
raise
def val(test_loader, model, epoch, save_path):
global best_mae, best_epoch
model.eval()
with torch.no_grad():
mae_sum = 0
for i in range(test_loader.size):
image, gt, depth, name, img_for_post = test_loader.load_data()
gt = np.asarray(gt, np.float32)
gt /= (gt.max() + 1e-8)
image = image.cuda()
depth = depth.cuda()
out= model(image, depth)
res = out
res = F.upsample(res, size=gt.shape, mode='bilinear', align_corners=False)
res = res.sigmoid().data.cpu().numpy().squeeze()
res = (res - res.min()) / (res.max() - res.min() + 1e-8)
mae_sum += np.sum(np.abs(res - gt)) * 1.0 / (gt.shape[0] * gt.shape[1])
mae = mae_sum / test_loader.size
print('Epoch: {} MAE: {} #### bestMAE: {} bestEpoch: {}'.format(epoch, mae, best_mae, best_epoch))
if epoch == 1:
best_mae = mae
else:
if mae < best_mae:
best_mae = mae
best_epoch = epoch
torch.save(model.state_dict(), save_path + 'SPNet_epoch_best.pth')
print('best epoch:{}'.format(epoch))
logging.info('#TEST#:Epoch:{} MAE:{} bestEpoch:{} bestMAE:{}'.format(epoch, mae, best_epoch, best_mae))
if __name__ == '__main__':
print("Start train...")
for epoch in range(1, opt.epoch):
train(train_loader, model, optimizer, epoch, save_path)
if epoch >= 80 or epoch ==1:
val(dataset_test, model, epoch, save_path)