-
Notifications
You must be signed in to change notification settings - Fork 4
/
main_pu.py
181 lines (120 loc) · 6.41 KB
/
main_pu.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
os.environ['CUDA_VISIBLE_DEVICES']='0'
import torch
import torch.nn as nn
from shapenet_small_dataset import mesh_pc_dataset
import numpy as np
import argparse
import model as model
#import loss
import logging
from glob import glob
#from pc_util import normalize_point_cloud, farthest_point_sample, group_points
from datetime import datetime
from tqdm import tqdm, trange
from torch.optim.lr_scheduler import CosineAnnealingLR
from pytorch3d.loss import chamfer_distance
from tensorboardX import SummaryWriter
#print(torch.cuda.is_available())
def log_string(out_str):
global LOG_FOUT
LOG_FOUT.write(out_str)
LOG_FOUT.flush()
if __name__=='__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--data_path', required=True, help='train or test')
parser.add_argument('--phase', default='train', help='train or test')
parser.add_argument('--gpu', default='0', help='which gpu to use')
parser.add_argument('--model', default='model_pugeo', help='Model for upsampling')
parser.add_argument('--min_num_point', type=int, default=3000, help='min Point Number')
parser.add_argument('--max_num_point', type=int, default=48000, help='max Point Number')
parser.add_argument('--batch_size', type=int, default=16, help='Batch Size during training')
parser.add_argument('--max_epoch', type=int, default=100, help='Epoch to run')
parser.add_argument('--learning_rate', type=float, default=0.001)
parser.add_argument('--min_lr', type=float, default=0.00001)
parser.add_argument('--pretrained', default='', help='Model stored')
parser.add_argument('--resume', type=bool, default=False, help='Number of points covered by patch')
arg = parser.parse_args()
arg.up_ratio=int(arg.max_num_point//arg.min_num_point)
arg.log_dir='log_x%d'%arg.up_ratio
try:
os.mkdir(arg.log_dir)
except:
pass
global LOG_FOUT
LOG_FOUT = open(os.path.join(arg.log_dir, 'log.txt'), 'w')
LOG_FOUT.write(str(datetime.now()) + '\n')
LOG_FOUT.write(os.path.abspath(__file__) + '\n')
LOG_FOUT.write(str(arg) + '\n')
dataset = mesh_pc_dataset(arg.data_path,mode='train',min_num_point=arg.min_num_point,max_num_point=arg.max_num_point,rot=False)
dataset_test = mesh_pc_dataset(arg.data_path,mode='val',min_num_point=arg.min_num_point,max_num_point=arg.max_num_point)
dataloader=torch.utils.data.DataLoader(dataset,batch_size=arg.batch_size,shuffle=True,drop_last=True,num_workers=16)
dataloader_test=torch.utils.data.DataLoader(dataset_test,batch_size=arg.batch_size//10,shuffle=False,drop_last=False,num_workers=16)
model=model.PUGeo(knn=20)
model = nn.DataParallel(model)
#model.load_state_dict(torch.load('log_x16_old/model_best.t7'))
model=model.cuda()
learnable_params = filter(lambda p: p.requires_grad, model.parameters())
current_lr=arg.learning_rate
optimizer=torch.optim.Adam(learnable_params,lr=current_lr)
scheduler = CosineAnnealingLR(optimizer, arg.max_epoch, eta_min=arg.min_lr)
loss_sum_dense_cd_test_best=1e10
writer = SummaryWriter(arg.log_dir)
global_step=0
for epoch in range(arg.max_epoch):
#scheduler.step()
loss_sum_all=[]
loss_sum_dense_cd = []
loss_sum_dense_normal = []
loss_sum_sparse_normal = []
model.train()
for data in tqdm(dataloader,desc='epoch %d train'%epoch):
global_step=global_step+1
input_sparse_xyz=data['sparse_pc']
gt_dense_xyz=data['dense_pc']
input_sparse_xyz = input_sparse_xyz.cuda()
gt_dense_xyz = gt_dense_xyz.cuda()
batch_size=gt_dense_xyz.size(0)
optimizer.zero_grad()
model.train()
output_dict=model(input_sparse_xyz,poisson=True,up_ratio=arg.up_ratio)
gen_dense_xyz=output_dict['dense_xyz']
loss_dense_cd=chamfer_distance(gen_dense_xyz.reshape(batch_size,-1,3),gt_dense_xyz.transpose(1,2))[0]
#loss_dense_cd,cd_idx1,cd_idx2=loss.cd_loss(gen_dense_xyz,gt_dense_xyz)
loss_all=100*loss_dense_cd
loss_all.backward()
optimizer.step()
loss_sum_all.append(loss_all.detach().cpu().numpy())
loss_sum_dense_cd.append(loss_dense_cd.detach().cpu().numpy())
if global_step%10==0:
writer.add_scalar('cd_loss', loss_dense_cd.detach().cpu().numpy().mean(), global_step)
loss_sum_all=np.array(loss_sum_all)
loss_sum_dense_cd=np.array(loss_sum_dense_cd)
log_string('epoch: %03d total loss: %0.7f, cd: %0.7f\n' % (
epoch, loss_sum_all.mean(), loss_sum_dense_cd.mean()
))
writer.add_scalar('Epoch_train_cd_loss', loss_sum_dense_cd.mean(), epoch)
torch.cuda.empty_cache()
loss_sum_dense_cd_test=[]
count_test=0
model.eval()
with torch.no_grad():
for data in tqdm(dataloader_test,desc='epoch %d test'%epoch):
count_test=count_test+input_sparse_xyz.size(0)
input_sparse_xyz=data['sparse_pc']
gt_dense_xyz=data['dense_pc']
input_sparse_xyz = input_sparse_xyz.cuda()
gt_dense_xyz = gt_dense_xyz.cuda()
batch_size=gt_dense_xyz.size(0)
output_dict=model(input_sparse_xyz,up_ratio=arg.up_ratio,poisson=True)
gen_dense_xyz=output_dict['dense_xyz']
#loss_dense_cd_test,cd_idx1,cd_idx2=loss.cd_loss(gen_dense_xyz,gt_dense_xyz)
loss_dense_cd_test=chamfer_distance(gen_dense_xyz.reshape(batch_size,-1,3),gt_dense_xyz.transpose(1,2))[0]
loss_sum_dense_cd_test.append(loss_dense_cd_test.detach().cpu().numpy())
loss_sum_dense_cd_test = np.asarray(loss_sum_dense_cd_test).mean()
writer.add_scalar('Epoch_test_cd_loss', loss_sum_dense_cd_test, epoch)
if loss_sum_dense_cd_test_best>loss_sum_dense_cd_test:
torch.save(model.state_dict(), os.path.join(arg.log_dir, 'model_best.t7'))
loss_sum_dense_cd_test_best=loss_sum_dense_cd_test
torch.save(model.state_dict(), os.path.join(arg.log_dir, 'model_last.t7'))
torch.cuda.empty_cache()