/
train.py
181 lines (143 loc) · 7.43 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
import os
import torch
import lpips
import torchvision
import open3d as o3d
import sys
import uuid
from tqdm import tqdm
from utils.loss_utils import l1_loss_w, ssim
from utils.general_utils import safe_state
from argparse import ArgumentParser, Namespace
from arguments import ModelParams, OptimizationParams, NetworkParams
from model.avatar_model import AvatarModel
from utils.general_utils import to_cuda, adjust_loss_weights
try:
from torch.utils.tensorboard import SummaryWriter
TENSORBOARD_FOUND = True
except ImportError:
TENSORBOARD_FOUND = False
def train(model, net, opt, saving_epochs, checkpoint_epochs):
tb_writer = prepare_output_and_logger(model)
avatarmodel = AvatarModel(model, net, opt, train=True)
loss_fn_vgg = lpips.LPIPS(net='alex').cuda()
train_loader = avatarmodel.getTrainDataloader()
iter_start = torch.cuda.Event(enable_timing = True)
iter_end = torch.cuda.Event(enable_timing = True)
first_iter = 0
epoch_start = 0
data_length = len(train_loader)
avatarmodel.training_setup()
if checkpoint_epochs:
avatarmodel.load(checkpoint_epochs[0])
epoch_start += checkpoint_epochs[0]
first_iter += epoch_start * data_length
if model.train_stage == 2:
avatarmodel.stage_load(model.stage1_out_path)
progress_bar = tqdm(range(first_iter, data_length * opt.epochs), desc="Training progress")
ema_loss_for_log = 0.0
for epoch in range(epoch_start + 1, opt.epochs + 1):
if model.train_stage ==1:
avatarmodel.net.train()
avatarmodel.pose.train()
avatarmodel.transl.train()
else:
avatarmodel.net.train()
avatarmodel.pose.eval()
avatarmodel.transl.eval()
avatarmodel.pose_encoder.train()
iter_start.record()
wdecay_rgl = adjust_loss_weights(opt.lambda_rgl, epoch, mode='decay', start=epoch_start, every=20)
for _, batch_data in enumerate(train_loader):
first_iter += 1
batch_data = to_cuda(batch_data, device=torch.device('cuda:0'))
gt_image = batch_data['original_image']
if model.train_stage ==1:
image, points, offset_loss, geo_loss, scale_loss = avatarmodel.train_stage1(batch_data, first_iter)
scale_loss = opt.lambda_scale * scale_loss
offset_loss = wdecay_rgl * offset_loss
Ll1 = (1.0 - opt.lambda_dssim) * l1_loss_w(image, gt_image)
ssim_loss = opt.lambda_dssim * (1.0 - ssim(image, gt_image))
loss = scale_loss + offset_loss + Ll1 + ssim_loss + geo_loss
else:
image, points, pose_loss, offset_loss, = avatarmodel.train_stage2(batch_data, first_iter)
offset_loss = wdecay_rgl * offset_loss
Ll1 = (1.0 - opt.lambda_dssim) * l1_loss_w(image, gt_image)
ssim_loss = opt.lambda_dssim * (1.0 - ssim(image, gt_image))
loss = offset_loss + Ll1 + ssim_loss + pose_loss * 10
if epoch > opt.lpips_start_iter:
vgg_loss = opt.lambda_lpips * loss_fn_vgg((image-0.5)*2, (gt_image- 0.5)*2).mean()
loss = loss + vgg_loss
avatarmodel.zero_grad(epoch)
loss.backward(retain_graph=True)
iter_end.record()
avatarmodel.step(epoch)
with torch.no_grad():
# Progress bar
ema_loss_for_log = 0.4 * loss.item() + 0.6 * ema_loss_for_log
if first_iter % 10 == 0:
progress_bar.set_postfix({"Loss": f"{ema_loss_for_log:.{7}f}"})
progress_bar.update(10)
if (first_iter-1) % opt.log_iter == 0:
save_poitns = points.clone().detach().cpu().numpy()
for i in range(save_poitns.shape[0]):
pcd = o3d.geometry.PointCloud()
pcd.points = o3d.utility.Vector3dVector(save_poitns[i])
o3d.io.write_point_cloud(os.path.join(model.model_path, 'log',"pred_%d.ply" % i) , pcd)
torchvision.utils.save_image(image, os.path.join(model.model_path, 'log', '{0:05d}_pred'.format(first_iter) + ".png"))
torchvision.utils.save_image(gt_image, os.path.join(model.model_path, 'log', '{0:05d}_gt'.format(first_iter) + ".png"))
if tb_writer:
tb_writer.add_scalar('train_loss_patches/l1_loss', Ll1.item(), first_iter)
tb_writer.add_scalar('train_loss_patches/total_loss', loss.item(), first_iter)
tb_writer.add_scalar('train_loss_patches/scale_loss', scale_loss.item(), first_iter)
tb_writer.add_scalar('train_loss_patches/offset_loss', offset_loss.item(), first_iter)
# tb_writer.add_scalar('train_loss_patches/aiap_loss', aiap_loss.item(), first_iter)
tb_writer.add_scalar('iter_time', iter_start.elapsed_time(iter_end), first_iter)
if model.train_stage ==1:
tb_writer.add_scalar('train_loss_patches/geo_loss', geo_loss.item(), first_iter)
else:
tb_writer.add_scalar('train_loss_patches/pose_loss', pose_loss.item(), first_iter)
if epoch > opt.lpips_start_iter:
tb_writer.add_scalar('train_loss_patches/vgg_loss', vgg_loss.item(), first_iter)
if (epoch > saving_epochs[0]) and epoch % model.save_epoch == 0:
print("\n[Epoch {}] Saving Model".format(epoch))
avatarmodel.save(epoch)
def prepare_output_and_logger(args):
if not args.model_path:
if os.getenv('OAR_JOB_ID'):
unique_str=os.getenv('OAR_JOB_ID')
else:
unique_str = str(uuid.uuid4())
args.model_path = os.path.join("./output/", unique_str[0:10])
# Set up output folder
print("Output folder: {}".format(args.model_path))
os.makedirs(args.model_path, exist_ok = True)
os.makedirs(os.path.join(args.model_path, 'log'), exist_ok = True)
with open(os.path.join(args.model_path, "cfg_args"), 'w') as cfg_log_f:
cfg_log_f.write(str(Namespace(**vars(args))))
# Create Tensorboard writer
tb_writer = None
if TENSORBOARD_FOUND:
tb_writer = SummaryWriter(args.model_path)
else:
print("Tensorboard not available: not logging progress")
return tb_writer
if __name__ == "__main__":
# Set up command line argument parser
parser = ArgumentParser(description="Training script parameters")
lp = ModelParams(parser)
op = OptimizationParams(parser)
np = NetworkParams(parser)
parser.add_argument('--debug_from', type=int, default=-1)
parser.add_argument('--detect_anomaly', action='store_true', default=False)
parser.add_argument("--save_epochs", nargs="+", type=int, default=[100])
parser.add_argument("--quiet", action="store_true")
parser.add_argument("--checkpoint_epochs", nargs="+", type=int, default=[])
parser.add_argument("--start_checkpoint", type=str, default = None)
args = parser.parse_args(sys.argv[1:])
args.save_epochs.append(args.epochs)
print("Optimizing " + args.model_path)
safe_state(args.quiet)
torch.autograd.set_detect_anomaly(args.detect_anomaly)
train(lp.extract(args), np.extract(args), op.extract(args), args.save_epochs, args.checkpoint_epochs)
print("\nTraining complete.")