-
Notifications
You must be signed in to change notification settings - Fork 13
/
train.py
310 lines (271 loc) · 13.4 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
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
from options import train_options
from train_utils import *
import logging
import os
import models_new as models
import torch
import torch.optim as optim
import torch.nn as nn
import torch.nn.functional as torch_f
from data.dataset import get_dataset
from data.datautils import ConcatDataloader
from traineval_util import data_dic, loss_func, save_2d_result,save_2d, mano_fitting, trans_proj, visualize, save_model, write_to_tb, dump
from utils.fh_utils import AverageMeter,EvalUtil
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
def TrainVal(mode_train, dat_name, epoch, train_loader, model, optimizer, requires, args, writer=None):
if mode_train:
model.train()
set_name = 'training'
else:
model.eval()
set_name = 'evaluation'
batch_time = AverageMeter()
end = time.time()
# Init output containers
evalutil = EvalUtil()
xyz_pred_list, verts_pred_list = list(), list()
op_xyz_pred_list, op_verts_pred_list = list(), list()
j2d_pred_ED_list, j2d_proj_ED_list, j2d_detect_ED_list = list(), list(), list()
for idx, (sample) in enumerate(train_loader):
# Get batch data
#import pdb; pdb.set_trace()
dat_name = sample['dataset']#check
examples = data_dic(sample, dat_name, set_name, args)
device = examples['imgs'].device
# Use some algorithm for prediction
if args.task == 'train' or args.task == 'test' or args.task == 'hm_train':
outputs = model(images=examples['imgs'], P=examples['Ps'], task=args.task, requires=requires)
# Projection transfer, project to 2D
outputs, xyz_pred_list, verts_pred_list = trans_proj(outputs, examples['Ks'], dat_name, xyz_pred_list, verts_pred_list)
#import pdb; pdb.set_trace()
# Compute loss function
loss_dic = loss_func(examples,outputs,dat_name,args)
# Compute and backward loss
loss = torch.zeros(1).float().to(device)
if dat_name == "RHD" and len(args.losses_rhd)>0:
loss_used = args.losses_rhd
elif dat_name == "FreiHand" and len(args.losses_frei)>0:
loss_used = args.losses_frei
else:
loss_used = args.losses
for loss_key in loss_used:
if loss_dic[loss_key]>0 and (not torch.isnan(loss_dic[loss_key]).sum()):
loss += loss_dic[loss_key].to(device)
#print(loss_key,loss_dic[loss_key],loss_dic[loss_key].device)
loss_dic['loss']=loss
#import pdb; pdb.set_trace()
if loss < 1e-10:
continue
if mode_train:
optimizer.zero_grad()
loss.backward()
optimizer.step()
op_outputs = None
else:
if args.test_refinement:
op_outputs, op_xyz_pred_list, op_verts_pred_list = mano_fitting(outputs,Ks=examples['Ks'], op_xyz_pred_list=op_xyz_pred_list, op_verts_pred_list=op_verts_pred_list,dat_name=dat_name, args=args)
else:
op_outputs = None
# save 2D results
if args.save_2d:
j2d_pred_ED_list, j2d_proj_ED_list, j2d_detect_ED_list = save_2d(examples, outputs, epoch, j2d_pred_ED_list, j2d_proj_ED_list, j2d_detect_ED_list, args)
# Save visualization and print informations
batch_time.update(time.time() - end)
visualize(mode_train,dat_name,epoch,idx,outputs,examples,args, op_outputs = op_outputs, writer=writer, writer_tag=set_name)
#import pdb; pdb.set_trace()
# print
if idx % args.print_freq == 0:
if optimizer is not None:
lr_current = optimizer.param_groups[0]['lr']
else:
lr_current = 0
print('Epoch: {0}\t'
'Iter: [{1}/{2}]\t'
'Time {batch_time.val:.3f}\t'
'Loss {loss:.5f}\t'
'datasat: {dataset:6}\t'
'lr {lr:.7f}\t'.format(epoch ,idx, len(train_loader),
batch_time=batch_time, loss=loss.data.item(), dataset=dat_name,
lr=lr_current))
print("Loss backward:\t",['{0}:{1:6f};'.format(loss_item,loss_data.sum()) for loss_item,loss_data in loss_dic.items() if (loss_item in loss_used and loss_data>1e-10)])
#print("Loss all:\t",['{0}:{1:6f};'.format(loss_item, loss_dic[loss_item].sum().data.item()) for loss_item in loss_dic])
#print("j3d loss:{0:.4f}; j2d loss:{1:.4f};shape loss:{2:.6f}; pose loss:{3:.6f}; render loss:{4:.6f}; sil loss:{5:.6f}; depth loss:{6:.5f}; render ssim loss:{7:.5f}; depth ssim loss:{8:.5f}; open j2d loss:{9:.5f}; mesh tex std:{10:.10f}; scale loss:{11:.5f}; bone direct loss:{12:.5f}; laplacian loss:{13:.6f}; hm loss:{14:.6f}; kp consistency loss:{15:.6f}; percep loss:{16:.6f}".format(joint_3d_loss.data.item(),joint_2d_loss.data.item(), shape_loss.data.item(),pose_loss.data.item(),texture_loss.data.item(), silhouette_loss.data.item(), depth_loss.data.item(), loss_ssim_tex.data.item(), loss_ssim_depth.data.item(), open_2dj_loss.data.item(), textures_reg.data.item(), mscale_loss.data.item(), open_bone_direc_loss.data.item(),triangle_loss.data.item(),hm_loss.data.item(),kp_cons_loss.data.item(),loss_percep.data.item()))
# write to tensorboard
if writer is not None:
with torch.no_grad():
write_to_tb(mode_train, writer, loss_dic, epoch, lr=optimizer.param_groups[0]['lr'])
#break
# dump results
if dat_name == 'FreiHand' or dat_name == 'HO3D':
if mode_train:
pred_out_path = os.path.join(args.pred_output,'train',str(epoch))
else:
pred_out_path = os.path.join(args.pred_output,'test',str(epoch))
if epoch%args.save_interval==0 and epoch>0:
os.makedirs(pred_out_path, exist_ok=True)
pred_out_path_0 = os.path.join(pred_out_path,'pred.json')
dump(pred_out_path_0, xyz_pred_list, verts_pred_list)
pred_out_op_path = os.path.join(pred_out_path,'pred_op.json')
dump(pred_out_op_path, op_xyz_pred_list, op_verts_pred_list)
if args.save_2d:
#import pdb;pdb.set_trace()
save_2d_result(j2d_pred_ED_list, j2d_proj_ED_list, j2d_detect_ED_list, args=args, epoch=epoch)
#break
def main(base_path, set_name=None, writer = None):
"""
Main eval loop: Iterates over all evaluation samples and saves the corresponding predictions.
"""
# default value
if set_name is None:
set_name = ['evaluation']
if 'training' in set_name:#set_name == 'training':
# initialize train datasets
train_loaders = []
if args.controlled_exp:
# Use subset of datasets so that final dataset size is constant
limit_size = int(args.controlled_size / len(args.train_datasets))
else:
limit_size = None
for dat_name in args.train_datasets:# iteration = min(dataset_len)/batch_size; go each dataset at a batchsize
if dat_name == 'FreiHand':
if len(args.train_queries_frei)>0:
train_queries = args.train_queries_frei
else:
train_queries = args.train_queries
base_path = args.freihand_base_path
elif dat_name == 'RHD':
if len(args.train_queries_rhd)>0:
train_queries = args.train_queries_rhd
else:
train_queries = args.train_queries
base_path = args.rhd_base_path
elif (dat_name == 'Obman') or (dat_name == 'Obman_hand'):
train_queries = args.train_queries
elif dat_name == 'HO3D':
if len(args.train_queries_ho3d)>0:
train_queries = args.train_queries_ho3d
else:
train_queries = args.train_queries
base_path = args.ho3d_base_path
train_dat = get_dataset(
dat_name,
'training',#set_name,
base_path,
queries = train_queries,
train = True,
limit_size=limit_size,
#transform=transforms.Compose([transforms.Rescale(256),transforms.ToTensor()]))
)
print("Training dataset size: {}".format(len(train_dat)))
# Initialize train dataloader
train_loader0 = torch.utils.data.DataLoader(
train_dat,
batch_size=args.train_batch,
shuffle=True,#check
num_workers=args.num_workers,
pin_memory=True,
drop_last=True,
)
train_loaders.append(train_loader0)
train_loader = ConcatDataloader(train_loaders)
#if 'evaluation' in set_name:
val_loaders = []
for dat_name_val in args.val_datasets:
if dat_name_val == 'FreiHand':
val_queries = args.val_queries
base_path = args.freihand_base_path
elif dat_name_val == 'RHD':
val_queries = args.val_queries
base_path = args.rhd_base_path
elif dat_name_val == 'HO3D':
val_queries = args.val_queries
base_path = args.ho3d_base_path
val_dat = get_dataset(
dat_name_val,
'evaluation',
base_path,
queries = val_queries,
train = False,
#transform=transforms.Compose([transforms.Rescale(256),transforms.ToTensor()]))
)
print("Validation dataset size: {}".format(len(val_dat)))
val_loader = torch.utils.data.DataLoader(
val_dat,
batch_size=args.val_batch,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
drop_last=False,
)
val_loaders.append(val_loader)
val_loader = ConcatDataloader(val_loaders)
#current_epoch = 0
if len(args.train_datasets) == 1:
dat_name = args.train_datasets[0]#dat_name
else:
dat_name = args.train_datasets
#losses = AverageMeter()
if 'training' in set_name:#set_name == 'training':
if args.optimizer == "Adam":
optimizer = optim.Adam(model.parameters(),lr=args.init_lr, betas=(0.9, 0.999), weight_decay=0)
if args.optimizer == "AdamW":
optimizer = optim.Adam(model.parameters(),lr=args.init_lr, betas=(0.9, 0.999), eps=1e-08, weight_decay=0.01, amsgrad=False)
scheduler = optim.lr_scheduler.MultiStepLR(optimizer, milestones=args.lr_steps, gamma=args.lr_gamma)
for epoch in range(1, args.total_epochs+1):
mode_train = True
requires = args.train_requires
args.train_batch = args.train_batch
TrainVal(mode_train, dat_name, epoch + current_epoch, train_loader, model, optimizer, requires, args, writer)
torch.cuda.empty_cache()
# save parameters
if (epoch + current_epoch) % args.save_interval == 0:
# test
mode_train = False
requires = args.test_requires
args.train_batch = args.val_batch
print('For test part:')
TrainVal(mode_train, dat_name_val, epoch + current_epoch, val_loader, model, optimizer, requires, args, writer)
torch.cuda.empty_cache()
save_model(model,optimizer,epoch,current_epoch, args)
scheduler.step()
elif 'evaluation' in set_name:#set_name == 'evaluation':
mode_train = False
requires = args.test_requires
optimizer = optim.Adam(model.parameters(),lr=args.init_lr, betas=(0.9, 0.999), weight_decay=0)#
#epoch = 0
#current_epoch = 0
#save_model(model,optimizer,epoch,current_epoch, args)
#import pdb; pdb.set_trace()
TrainVal(mode_train, dat_name_val, current_epoch, val_loader, model, None, requires, args, writer)
print("Finish write prediction. Good luck!")
print("Done!")
if __name__ == '__main__':
args = train_options.parse()
if args.config_json is not None:
with open(args.config_json, "r") as f:
json_dic = json.load(f)
for parse_key, parse_value in json_dic.items():
vars(args)[parse_key] = parse_value
args = train_options.make_output_dir(args)
#import pdb; pdb.set_trace()#args.update(args.__dic__)
if args.is_write_tb:
writer = SummaryWriter(log_dir=os.path.abspath(os.path.dirname(os.path.dirname(__file__)))+args.writer_topic+datetime.now().strftime("%Y%m%d-%H%M%S"))
print(datetime.now().strftime("%Y%m%d-%H%M%S"))
else:
writer = None
logging.basicConfig(format='%(asctime)s %(message)s', datefmt='%Y/%m/%d %H:%M:%S', filename=os.path.join(args.base_output_dir, 'train.log'), level=logging.INFO)
logging.info("=====================================================")
model = models.Model('/code/TextureHand/examples/data/obj/sphere/sphere_642.obj', args=args)
model, current_epoch = load_model(model, args)
model = nn.DataParallel(model.cuda())
# Optionally freeze parts of the network
freeze_model_modules(model, args)
# call with a predictor function
main(
args.base_path,
#args.out,
set_name=args.mode,
writer = writer,
)
if writer is not None:
writer.close()