-
-
Notifications
You must be signed in to change notification settings - Fork 0
/
train4.py
411 lines (360 loc) · 18.3 KB
/
train4.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
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
###在动态数据集上加东西 图片一般是(h, w, 3)
import imp
import os, sys
from tkinter import image_names
from opt import config_parser
import torch
from collections import defaultdict
import random
from torch.utils.data import DataLoader
from datasets import dataset_dict
import pdb
# models
from models.nerf import Embedding, NeRF, LatentCode
from models.rendering import render_rays, render_rays1, render_rays2
from models.model_nerv import *
# optimizer, scheduler, visualization, NeRV utils
from utils import *
from utils.NeRV import *
# losses
from losses import loss_dict
import imageio
# metrics
from metrics import *
from torchvision import transforms
# pytorch-lightning
from pytorch_lightning.callbacks import ModelCheckpoint
from pytorch_lightning import LightningModule, Trainer
from pytorch_lightning.loggers import TestTubeLogger
from pytorch_lightning.callbacks.early_stopping import EarlyStopping
device = torch.device("cuda:2" if torch.cuda.is_available() else "cpu")
os.environ['CUDA_VISIBLE_DEVICES'] = "0, 1, 2 ,3"
# os.environ['CUDA_VISIBLE_DEVICES'] = "1"
# device = torch.device("cuda")
class NeRFSystem(LightningModule):
def __init__(self, args):
super(NeRFSystem, self).__init__()
self.args = args
self.idx = 0
self.loss = loss_dict[args.loss_type]()
self.NeRV_ckpt = args.weight + args.view_choose + args.NeRV_model
self.NeRV_ckpt = []
for view in range(18):
view_choose = f'/cam{view:02d}'
self.NeRV_ckpt += [args.weight + view_choose + args.NeRV_model]
self.NeRV_list = []
# self.embedding_xyz = Embedding(3, 10) # 10 is the default number
#NeRV对t进行pe编码
self.PE = PositionalEncoding(args.embed)
args.embed_length = self.PE.embed_length
self.embedding_xyz = Embedding(3, 10)
self.embedding_dir = Embedding(3, 4) # 4 is the default number
self.embeddings = [self.embedding_xyz, self.embedding_dir]
self.list_log = []
self.img_wh = self.args.img_wh
self.image_shape = torch.zeros(self.img_wh[1], self.img_wh[0], 3)
args.single_res = True
# self.model_NeRV_cam00 = Generator(embed_length=args.embed_length, stem_dim_num=args.stem_dim_num, fc_hw_dim=args.fc_hw_dim, expansion=args.expansion,
# num_blocks=args.num_blocks, norm=args.norm, act=args.act, bias = True, reduction=args.reduction, conv_type=args.conv_type,
# stride_list=args.strides, sin_res=args.single_res, lower_width=args.lower_width, sigmoid=args.sigmoid)
# NeRVs = [Generator(embed_length=args.embed_length, stem_dim_num=args.stem_dim_num, fc_hw_dim=args.fc_hw_dim, expansion=args.expansion,
# num_blocks=args.num_blocks, norm=args.norm, act=args.act, bias = True, reduction=args.reduction, conv_type=args.conv_type,
# stride_list=args.strides, sin_res=args.single_res, lower_width=args.lower_width, sigmoid=args.sigmoid) for i in range(18)]
# self.NeRVs = nn.ModuleList(NeRVs)
# for name, p in self.NeRVs.named_parameters():
# print(name)
# for name, module in self.NeRVs.named_modules():
# if name == '0':
# print(module)
self.model_NeRV = {}
for x in range(18):
self.model_NeRV[f'self.model_NeRV_cam{x:02d}'] = Generator(embed_length=args.embed_length, stem_dim_num=args.stem_dim_num, fc_hw_dim=args.fc_hw_dim, expansion=args.expansion,
num_blocks=args.num_blocks, norm=args.norm, act=args.act, bias = True, reduction=args.reduction, conv_type=args.conv_type,
stride_list=args.strides, sin_res=args.single_res, lower_width=args.lower_width, sigmoid=args.sigmoid).cuda()
# for i in range(1, 18):
# f'self.model_NeRV_cam_{i:02d}' =
self.NeRV_keys = []
# for i in range(len(self.NeRV_ckpt)):
for i, key in enumerate(self.model_NeRV.keys()):
self.NeRV_keys += [key]
# model = self.model_NeRV[key]
# exec('model_{} = {}'.format(i, self.model_NeRV_cam00))
# setattr(self, f"model_{i:02d}", self.model_NeRV_cam00)
# model_0 = self.model_NeRV_cam00
if self.NeRV_ckpt[i] != 'None':
# model = self.model_NeRV[f'self.model_NeRV_cam{i:02d}']
print("=> loading checkpoint '{}'".format(self.NeRV_ckpt[i]))
checkpoint_path = self.NeRV_ckpt[i]
checkpoint = torch.load(checkpoint_path, map_location='cpu')
orig_ckt = checkpoint['state_dict']
new_ckt={k.replace('blocks.0.',''):v for k,v in orig_ckt.items()}
self.model_NeRV[key].to(device)
if 'module' in list(orig_ckt.keys())[0] and not hasattr(self.model_NeRV[key], 'module'):
new_ckt={k.replace('module.',''):v for k,v in new_ckt.items()}
self.model_NeRV[key].load_state_dict(new_ckt)
elif 'module' not in list(orig_ckt.keys())[0] and hasattr(self.model_NeRV[key], 'module'):
self.model_NeRV[key].module.load_state_dict(new_ckt)
else:
self.model_NeRV[key].load_state_dict(new_ckt)
print("=> loaded checkpoint '{}' (epoch {})".format(args.weight, checkpoint['epoch']))
self.nerf_coarse = NeRF()
load_ckpt(self.nerf_coarse, args.ckpt_path, model_name='nerf_coarse')
self.models = [self.nerf_coarse]
if args.N_importance > 0:
self.nerf_fine = NeRF()
load_ckpt(self.nerf_fine, args.ckpt_path, model_name='nerf_fine')
self.models += [self.nerf_fine]
self.LatentCode = LatentCode()
load_ckpt(self.LatentCode, args.ckpt_path, model_name='LatentCode')
self.models += [self.LatentCode]
self.t_normalize = 0
self.flag_image_num = 0
self.flag_epoch1 = 0
self.flag_epoch2 = 0
self.list_all = list(range(0, 480*640))
self.list_all_1 = self.list_all[:]
random.shuffle(self.list_all_1)
self.num = 0
def decode_batch(self, batch):
rays = batch['rays'] # (B, 9)
rgbs = batch['rgbs'] # (B, 3)
image_t = batch['image_t']
flag_Interpolation = batch['flag']
view = batch['view']
pixel_choose = batch['pixel_choose']
# image_t = batch['time']
return rays, rgbs, image_t, flag_Interpolation, view, pixel_choose
def unpreprocess(self, data, shape=(1,3,1,1)):
# to unnormalize image for visualization
# data N V C H W
device = data.device
mean = torch.tensor([-0.485 / 0.229, -0.456 / 0.224, -0.406 / 0.225]).view(*shape).to(device)
std = torch.tensor([1 / 0.229, 1 / 0.224, 1 / 0.225]).view(*shape).to(device)
return (data - mean) / std
def forward(self, rays, t_normalize = 0):
"""Do batched inference on rays using chunk."""
B = rays.shape[0] #160000
results = defaultdict(list)
for i in range(0, B, self.args.chunk):
rendered_ray_chunks = \
render_rays2(self.models,
self.embeddings,
rays[i:i+self.args.chunk], #[32768, 8]
self.args.N_samples,
self.args.use_disp,
self.args.perturb,
self.args.noise_std,
self.args.N_importance,
self.args.chunk, # chunk size is effective in val mode 32768
self.train_dataset.white_back,
t_normalize = t_normalize,
test_time=False
)
for k, v in rendered_ray_chunks.items():
results[k] += [v] #k 'rgb_coarse' v为数值
for k, v in results.items():
results[k] = torch.cat(v, 0)
return results
def prepare_data(self):
dataset = dataset_dict[self.args.dataset_name]
# self.train_dataset = dataset(split='train', **kwargs)
# self.val_dataset = dataset(split='val', **kwargs)
train_dir = val_dir = self.args.root_dir
self.train_dataset = dataset(root_dir=train_dir, split='train', max_len=-1)
self.val_dataset = dataset(root_dir=val_dir, split='val', max_len=10)
def configure_optimizers(self):
self.optimizer = get_optimizer(self.args, self.models)
scheduler = get_scheduler(self.args, self.optimizer)
return [self.optimizer], [scheduler]
def train_dataloader(self):
return DataLoader(self.train_dataset,
shuffle=True,
num_workers=0,
# batch_size=self.args.batch_size,
batch_size=1,
pin_memory=True)
def val_dataloader(self):
return DataLoader(self.val_dataset,
shuffle=False,
num_workers=0,
batch_size=1, # validate one image (H*W rays) at a time
pin_memory=True)
def training_step(self, batch, batch_nb):
log = {'lr': get_learning_rate(self.optimizer)}
rays, rgbs, image_t, flag_Interpolation, view, pixel_choose = self.decode_batch(batch)
if flag_Interpolation == True: #gt要从NeRV网络中得到
with torch.no_grad():
loss_weight = 0.01
model_NeRV = self.model_NeRV[self.NeRV_keys[view]]
embed_t = self.PE(image_t)
embed_t = embed_t.cuda(non_blocking=True)
rgbs = model_NeRV(embed_t.float())[0].squeeze().reshape(3, -1).permute(1, 0)
rgbs = rgbs[pixel_choose]
if flag_Interpolation == False:
# rgbs = rgbs.squeeze()
model_NeRV = self.model_NeRV[self.NeRV_keys[view]]
embed_t = self.PE(image_t)
embed_t = embed_t.cuda(non_blocking=True)
rgbs = model_NeRV(embed_t.float())[0].squeeze().reshape(3, -1).permute(1, 0)
rgbs = rgbs[pixel_choose]
loss_weight = 1
# image_t = rays[:, -1].unsqueeze(1)
# t_num1 = torch.tensor(30.0)
# t_normalize = 2*scene_t/(t_num1-1)-1 #0
# t_normalize = 2*image_t.type(torch.FloatTensor)/(t_num1.type(torch.FloatTensor))-1
# t_normalize = 0
# list = random.sample(range(0,2704 * 2028),1024)
t_normalize = image_t
rays = rays.squeeze()[:, :8]
# rgbs = rgbs[self.point_list]
# rays = rays[self.point_list]
results = self(rays.float(), t_normalize = t_normalize) #[1024,3]
log['train/loss'] = loss = self.loss(results, rgbs) * loss_weight
typ = 'fine' if 'rgb_fine' in results else 'coarse'
# w, h = self.hparams.img_wh
# img_pred = results['rgb_fine'].view(h, w, 3).cpu().numpy()
# img_pred_ = (img_pred*255).astype(np.uint8)
# dir_name = '/home/liufengyi/test/nerf_pl-master/nerf_pl-master'
# i = 0
# imageio.imwrite(os.path.join(dir_name, f'{i:03d}.png'), img_pred_)
with torch.no_grad():
psnr_ = psnr(results[f'rgb_{typ}'], rgbs)
log['train/psnr'] = psnr_
self.log('train/loss', loss.item(), prog_bar=True)
self.log('train/psnr', psnr_.item(), prog_bar=True)
# self.logger.experiment.add_images
return {'loss': loss}
return {'loss': loss,
'progress_bar': {'train/psnr': psnr_},
'log': log
}
def validation_step(self, batch, batch_nb):
self.idx += 1
if self.idx == 1:
self.idx_gpus = 0
else:
self.idx_gpus = (self.idx - 2)//8 + 1
# else:
# self.idx_gpus = (self.idx-2)//4 + 1
# self.idx_gpus = self.idx//8
rays, rgbs, image_t, flag_Interpolation, view, pixel_choose = self.decode_batch(batch)
time = int(image_t*30)
# # print("val_t:",image_t)
# rays = rays.squeeze() # (H*W, 3) [160000, 8]
# rgbs = rgbs.squeeze() # (H*W, 3)
# t_num1 = torch.tensor(30.0)
# image_t = image_t*torch.ones_like(rays[:, :1])
# t_normalize = 2*image_t.type(torch.FloatTensor)/(t_num1.type(torch.FloatTensor))-1
rays = rays.squeeze() # (H*W, 3) [160000, 8]
rgbs = rgbs.squeeze() # (H*W, 3)
t_normalize = image_t
results = self(rays.float(), t_normalize = t_normalize) #[160000,3]
# results = self(rays.float())
log = {'val_loss': self.loss(results, rgbs)}
typ = 'fine' if 'rgb_fine' in results else 'coarse'
log['val_psnr'] = psnr(results[f'rgb_{typ}'], rgbs)
# img = results['rgb_coarse'].reshape(400,400,3).permute(2,0,1).cpu()
img = results[f'rgb_{typ}'].reshape(*self.image_shape.shape).permute(2,0,1).cpu()
# img1 = rgbs.reshape(400,400,3).permute(2,0,1).cpu()
img1 = rgbs.reshape(*self.image_shape.shape).permute(2,0,1).cpu()
# img = img.cuda()
# img1 = img1.cuda()
# img = self.unpreprocess(img).squeeze().cpu()
# img1 = self.unpreprocess(img1).squeeze().cpu()
# toPIL = transforms.ToPILImage()
# pic = toPIL(img)
# # pic.save(f'pre_{batch_nb}.jpg')
# pic = toPIL(img1)
# pic.save(f'gt_{batch_nb}.jpg')
img1 = img1.unsqueeze(0)
img = img.unsqueeze(0)
img_vis = torch.cat((img1,img),dim=0).permute(2,0,3,1).reshape(img1.shape[2],-1,3).numpy()
os.makedirs(f'/data1/liufengyi/get_results/non_synchronized_NeRF2/runs_new_try1/{self.args.exp_name}/{self.args.exp_name}/',exist_ok=True)
imageio.imwrite(f'/data1/liufengyi/get_results/non_synchronized_NeRF2/runs_new_try1/{self.args.exp_name}/{self.args.exp_name}/{time:02d}_{self.idx_gpus:02d}.png', (img_vis*255).astype('uint8'))
# if batch_nb == 0:
# W, H = self.args.img_wh
# img = results[f'rgb_{typ}'].view(H, W, 3).cpu()
# img = img.permute(2, 0, 1) # (3, H, W)
# img_gt = rgbs.view(H, W, 3).permute(2, 0, 1).cpu() # (3, H, W)
# depth = visualize_depth(results[f'depth_{typ}'].view(H, W)) # (3, H, W)
# stack = torch.stack([img_gt, img, depth]) # (3, 3, H, W)
# self.logger.experiment.add_images('val/GT_pred_depth',
#
# stack, self.global_step)
return log
def validation_epoch_end(self, outputs):
self.flag_epoch1 += 1
self.point_list = self.list_all_1[(self.flag_epoch1 - 1) * 1024 : self.flag_epoch1 * 1024]
mean_loss = torch.stack([x['val_loss'] for x in outputs]).mean()
mean_psnr = torch.stack([x['val_psnr'] for x in outputs]).mean()
self.log('val_loss', mean_loss.item(), prog_bar=True)
self.log('val_psnr', mean_psnr.item(), prog_bar=True)
return
return {'progress_bar': {'val_loss': mean_loss,
'val_psnr': mean_psnr},
'log': {'val/loss': mean_loss,
'val/psnr': mean_psnr}
}
def save_ckpt(self, name='final'):
save_dir = f'/data1/liufengyi/get_results/non_synchronized_NeRF2/runs_new/{self.args.exp_name}/ckpts/'
os.makedirs(save_dir, exist_ok=True)
path = f'{save_dir}/{name}.tar'
ckpt = {
# 'global_step': self.global_step,
# 'network_fn_state_dict': self.render_kwargs_train['network_fn'].state_dict(),
# 'network_mvs_state_dict': self.MVSNet.state_dict()}
'nerf_coarse_state_dict' :self.models[0].state_dict(),
'nerf_fine_state_dict' :self.models[1].state_dict(),
'latent_code': self.models[2].state_dict()}
# if self.render_kwargs_train['network_fine'] is not None:
# ckpt['network_fine_state_dict'] = self.render_kwargs_train['network_fine'].state_dict()
torch.save(ckpt, path)
print('Saved checkpoints at', path)
if __name__ == '__main__':
args = config_parser()
system = NeRFSystem(args)
a = os.path.join(f'/data1/liufengyi/get_results/non_synchronized_NeRF2/runs_new/{args.exp_name}/ckpts/','{epoch:02d}')
dirpath = f'/data1/liufengyi/get_results/non_synchronized_NeRF2/runs_new/{args.exp_name}/ckpts/'
# filename = '{epoch:02d}'
filename = '{epoch:02d}-{val_loss:.3f}'
# early_stop_callback = (
# EarlyStopping(
# monitor = 'val/loss_mean',
# patience = 15,
# mode = 'min')
# )
checkpoint_callback = ModelCheckpoint(dirpath = dirpath,
filename = filename,
monitor='val_psnr',
mode='max',
save_top_k=5,)
# auto_insert_metric_name=False)
logger = TestTubeLogger(
save_dir="/data1/liufengyi/get_results/non_synchronized_NeRF2/logs",
name=args.exp_name,
debug=False,
create_git_tag=False
)
trainer = Trainer(max_epochs=args.num_epochs,
# checkpoint_callback=checkpoint_callback,
# callbacks=[checkpoint_callback, early_stop_callback],
callbacks=[checkpoint_callback],
# resume_from_checkpoint=args.ckpt_path,
logger=logger,
# early_stop_callback=None,
weights_summary=None,
progress_bar_refresh_rate=1,
# gpus=args.num_gpus,
gpus=[2],
distributed_backend='ddp' if args.num_gpus>1 else None,
num_sanity_val_steps = 1, #训练之前进行校验
# check_val_every_n_epoch = 1, #一个epoch校验一次
val_check_interval=0.25, #0.1个epoch校验一次
precision=16,
benchmark=True,
log_every_n_steps = 50, ) #每隔1次迭代记录一下logger
# profiler=args.num_gpus==1)
# pdb.set_trace()
trainer.fit(system)
system.save_ckpt()
torch.cuda.empty_cache()