-
Notifications
You must be signed in to change notification settings - Fork 9
/
train_point.py
188 lines (155 loc) · 6.5 KB
/
train_point.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
import torch
import torch.nn.functional as F
import numpy as np
from tqdm import tqdm
from model_cl import *
from datasets import Dataset3D
from tensorboardX import SummaryWriter
def train_step(train_data, weight_model, robust_estimator, data_type, prob_type=0, dev='cuda'):
weight_model.to(data_type)
# fetch the points, ground truth extrinsic and intrinsic matrices
correspondences, gt_pose = train_data['correspondences'].to(dev, data_type), \
train_data['gt_pose'].to(dev, data_type)
# 1. importance score prediction
weights = weight_model(correspondences.transpose(-1, -2)[:, :, :, None])
# import pdb; pdb.set_trace()
# 2. ransac
loss_back = 0
for i, pair in enumerate(correspondences[:, :, :6]):
Es, loss, avg_loss, _ = robust_estimator(
pair,
weights[i],
gt_pose[i]
)
loss_back += avg_loss
return loss_back/correspondences.shape[0]
def train(
model,
estimator,
train_loader,
valid_loader,
opt
):
# the name of the folder we save models, logs
saved_file = create_session_string(
"train",
opt.sampler,
opt.epochs,
opt.fmat,
opt.nfeatures,
opt.snn,
opt.session,
opt.w0,
opt.w1,
opt.w2,
opt.threshold
)
writer = SummaryWriter('results/point/' + saved_file + '/vision', comment="model_vis")
optimizer = torch.optim.Adam(model.parameters(), lr=opt.learning_rate)
valid_loader_iter = iter(valid_loader)
# save the losses to npy file
train_losses = []
valid_losses = []
if opt.precision == 2:
data_type = torch.float64
elif opt.precision == 0:
data_type = torch.float16
else:
data_type = torch.float32
# start epoch
for epoch in range(opt.epochs):
# each step
for idx, train_data in enumerate(tqdm(train_loader)):
model.train()
# one step
optimizer.zero_grad()
train_loss = train_step(train_data, model, estimator, data_type, prob_type=opt.prob, dev=opt.device)
train_loss.retain_grad()
# gradient calculation, ready for back propagation
if torch.isnan(train_loss):
print("pls check, there is nan value in loss!", train_loss)
continue
try:
train_loss.backward()
print("successfully back-propagation", train_loss)
except Exception as e:
print("we have trouble with back-propagation, pls check!", e)
continue
if torch.isnan(train_loss.grad):
print("pls check, there is nan value in the gradient of loss!", train_loss.grad)
continue
train_losses.append(train_loss.cpu().detach().numpy())
# for vision
writer.add_scalar('train_loss', train_loss, global_step=epoch*len(train_loader)+idx)
# add gradient clipping after backward to avoid gradient exploding
torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm=5)
# check if the gradients of the training parameters contain nan values
nans = sum([torch.isnan(param.grad).any() for param in list(model.parameters()) if param.grad is not None])
if nans != 0:
print("parameter gradients includes {} nan values".format(nans))
continue
optimizer.step()
# check check if the training parameters contain nan values
nan_num = sum([torch.isnan(param).any() for param in optimizer.param_groups[0]['params']])
if nan_num != 0:
print("parameters includes {} nan values".format(nan_num))
continue
torch.save(model.state_dict(), 'results/point/' + saved_file + '/model' + str(epoch) + '.net')
print("_______________________________________________________")
# validation
with torch.no_grad():
model.eval()
try:
valid_data = next(valid_loader_iter)
except StopIteration:
pass
valid_loss = train_step(valid_data, model, estimator, data_type, prob_type=opt.prob, dev=opt.device)
valid_losses.append(valid_loss)
writer.add_scalar('valid_loss', valid_loss, global_step=epoch * len(train_loader) + idx)
writer.flush()
print('Step: {:02d}| Train loss: {:.4f}| Validation loss: {:.4f}'.format(
epoch*len(train_loader)+idx,
train_loss,
valid_loss
), '\n')
np.save('results/point/' + saved_file + '/' + 'loss_record.npy', (train_losses, valid_losses))
if __name__ == '__main__':
# Parse the parameters
parser = create_parser(
description="Generalized Differentiable RANSAC.")
config = parser.parse_args()
# check if gpu device is available
config.device = torch.device('cuda:0' if torch.cuda.is_available() and config.device != 'cpu' else 'cpu')
print(f"Running on {config.device}")
train_model = CLNet().to(config.device)
robust_estimator = RANSACLayer3D(config)
# use the pretrained model to initialize the weights if provided.
if config.model is not None:
train_model.load_state_dict(torch.load(config.model))
else:
train_model.apply(init_weights)
train_model.train()
# collect dataset list
train_scenes = os.listdir(config.data_path)
train_folders = [config.data_path + '/' + i + '/' for i in train_scenes]
train_dataset = Dataset3D(train_folders)
v_folders = [config.data_path.replace('train', 'val') + '/' + i + '/' for i in os.listdir(config.data_path.replace('train', 'val'))]
v_dataset = Dataset3D(v_folders)
print("\n=== BATCH MODE: Training and validation on", len(train_scenes), len(v_folders), "datasets. =================")
train_data_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=config.batch_size,
num_workers=config.num_workers,
pin_memory=True,
shuffle=True
)
print(f'Loading training data: {len(train_dataset)} image pairs.')
valid_data_loader = torch.utils.data.DataLoader(
v_dataset,
batch_size=config.batch_size,
num_workers=config.num_workers,
pin_memory=True,
shuffle=True
)
print(f'Loading validation data: {len(v_dataset)} image pairs.')
train(train_model, robust_estimator, train_data_loader, valid_data_loader, config)