-
Notifications
You must be signed in to change notification settings - Fork 1
/
train.py
567 lines (483 loc) · 18.6 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
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
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
import json
import os
import random
import sys
import time
import torch
import numpy as np
import hydra
import shutil
import pytz
from datetime import datetime
from omegaconf import DictConfig, OmegaConf
from collections import defaultdict
from torch.utils.tensorboard import SummaryWriter
from tqdm.auto import tqdm
from opt import config_parser
from models.tensoRF import TensorVMSplit, TensorCP
from dataLoader import dataset_dict
from renderer import (
OctreeRender_trilinear_fast,
evaluation,
create_gif,
evaluation_path,
save_rendered_image_per_train
)
from loss import (
TVLoss,
PSNRs_calculate
)
from utils import (
convert_sdf_samples_to_ply,
N_to_reso,
cal_n_samples,
get_free_mask,
)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
renderer = OctreeRender_trilinear_fast
class SimpleSampler:
def __init__(self, total, batch):
self.total = total
self.batch = batch
self.curr = total
self.ids = None
def nextids(self):
self.curr += self.batch
if self.curr + self.batch > self.total:
self.ids = torch.LongTensor(np.random.permutation(self.total))
self.curr = 0
return self.ids[self.curr : self.curr + self.batch]
@torch.no_grad()
@torch.no_grad()
def export_mesh(args, ckpt_path):
ckpt = torch.load(ckpt_path, map_location=device)
kwargs = ckpt['kwargs']
kwargs.update({'device': device})
tensorf = eval(args.model_name)(args, **kwargs)
tensorf.load(ckpt)
alpha,_ = tensorf.getDenseAlpha()
convert_sdf_samples_to_ply(
alpha.cpu(),
f'{ckpt_path[:-3]}.ply',
bbox=tensorf.aabb.cpu(),
level=0.005
)
@torch.no_grad()
def render_test(args):
# init dataset
dataset = dataset_dict[args.dataset_name]
test_dataset = dataset(
args.datadir,
split="test",
downsample=args.downsample_train,
is_stack=True,
)
white_bg = test_dataset.white_bg
ndc_ray = args.ndc_ray
if not os.path.exists(args.ckpt):
print("the ckpt path does not exists!!")
return
ckpt = torch.load(args.ckpt, map_location=device)
kwargs = ckpt["kwargs"]
kwargs.update({"device": device})
occ_grid = None
if args.occ_grid_reso > 0:
occ_grid = nerfacc.OccGridEstimator(
roi_aabb=ckpt["state_dict"]["occGrid.aabbs"][0],
resolution=args.occ_grid_reso,
).to(device)
tensorf = eval(args.model_name)(**kwargs, occGrid=occ_grid)
tensorf.load(ckpt)
logfolder = os.path.dirname(args.ckpt)
if args.render_train:
os.makedirs(f"{logfolder}/imgs_train_all", exist_ok=True)
train_dataset = dataset(
args.datadir,
split="train",
downsample=args.downsample_train,
is_stack=True,
)
PSNRs_test = evaluation(
train_dataset,
tensorf,
args,
renderer,
f"{logfolder}/imgs_train_all/",
N_vis=-1,
N_samples=-1,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device,
)
print(
f"======> {args.expname} train all psnr: {np.mean(PSNRs_test)} <========================"
)
if args.render_test:
os.makedirs(f"{logfolder}/imgs_test_all", exist_ok=True)
PSNRs_test = evaluation(
test_dataset,
tensorf,
args,
renderer,
f"{logfolder}/imgs_test_all/",
N_vis=-1,
N_samples=-1,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device,
)
print(
f"======> {args.expname} test all psnr: {np.mean(PSNRs_test)} <========================"
)
if args.render_path:
c2ws = test_dataset.render_path
os.makedirs(f"{logfolder}/imgs_path_all", exist_ok=True)
print(f"{logfolder}/imgs_path_all")
evaluation_path(
test_dataset,
tensorf,
c2ws,
renderer,
f"{logfolder}/imgs_path_all/",
N_vis=-1,
N_samples=-1,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device,
)
@hydra.main(version_base=None)
def reconstruction(args: DictConfig):
# ==> Dataset
# ================================
dataset = dataset_dict[args.dataset_name]
train_dataset = dataset(args.datadir, split='train', downsample=args.downsample_train, num_images=OmegaConf.to_object(args.train_images))
test_dataset = dataset(args.datadir, split='test', downsample=args.downsample_train, num_images=OmegaConf.to_object(args.test_images), is_stack=True)
# final_test_dataset = dataset(args.datadir, split='test', downsample=args.downsample_train, is_stack=True, num_images=args.N_train_imgs)
train_gift_data = dataset(args.datadir, split='train', downsample=args.downsample_train, num_images=[26], is_stack=True)
test_gift_data = dataset(args.datadir, split='test', downsample=args.downsample_train, num_images=[26], is_stack=True)
white_bg = train_dataset.white_bg
near_far = train_dataset.near_far
ndc_ray = args.ndc_ray
# ==> Init resolution
# ================================
upsamp_list = args.upsamp_list
update_AlphaMask_list = args.update_AlphaMask_list
n_lamb_sigma = args.n_lamb_sigma
n_lamb_sh = args.n_lamb_sh
# ==> Init log folder
# ================================
timezone = pytz.timezone('Asia/Ho_Chi_Minh')
current_time = datetime.now(timezone)
logfolder = f'{args.basedir}/{current_time.strftime("%Y-%m-%d")}/{args.expname}'
if args.overwrt and os.path.exists(logfolder): shutil.rmtree(logfolder)
os.makedirs(logfolder, exist_ok=True)
os.makedirs(f"{logfolder}/imgs_vis", exist_ok=True)
os.makedirs(f"{logfolder}/imgs_rgba", exist_ok=True)
os.makedirs(f"{logfolder}/rgba", exist_ok=True)
summary_writer = SummaryWriter(logfolder)
# ==> Init parameters
# ================================
aabb = train_dataset.scene_bbox.to(device)
gridSize = N_to_reso(args.N_voxel_init, aabb)
nSamples = min(1e6, cal_n_samples(gridSize, args.step_ratio))
N_voxel_list = (
torch.round(
torch.exp(
torch.linspace(np.log(args.N_voxel_init), np.log(args.N_voxel_final), len(upsamp_list) + 1)
)
).long()
).tolist()[1:]
# ==> Load checkpoint
# ================================
if args.ckpt_path is not None:
ckpt = torch.load(args.ckpt_path, map_location=device)
kwargs = ckpt["kwargs"]
kwargs.update({"device": device})
tensorf = eval(args.model_name)(**kwargs)
tensorf.load(ckpt)
else:
tensorf = eval(args.model_name)(
args={
'step_ratio': args.step_ratio,
'fea2denseAct': args.fea2denseAct,
'density_n_comp': args.n_lamb_sigma,
'app_n_comp': args.n_lamb_sh,
'app_dim': args.data_dim_color,
'density_shift': args.density_shift,
'distance_scale': args.distance_scale,
'alphaMask_thres': args.alphaMask_thres,
'shadingMode': args.shadingMode,
'pos_pe': args.pos_pe,
'view_pe': args.view_pe,
'fea_pe': args.fea_pe,
'featureC': args.featureC
},
aabb=aabb,
gridSize=gridSize,
device=device,
near_far=train_dataset.near_far
)
torch.cuda.empty_cache()
print(f"initial TV_weight density: {args.TV_weight_density} appearance: {args.TV_weight_app}")
# ==> Loss init
# ================================
Ortho_reg_weight = args.Ortho_weight
print("initial Ortho_reg_weight", Ortho_reg_weight)
L1_reg_weight = args.L1_weight_inital
print("initial L1_reg_weight", L1_reg_weight)
TV_weight_density, TV_weight_app = args.TV_weight_density, args.TV_weight_app
tvreg = TVLoss()
# ==> Optimzier and Learning rate
# ================================
if args.lr_decay_iters > 0:
lr_factor = args.lr_decay_target_ratio ** (1 / args.lr_decay_iters)
else:
lr_decay_iters = args.n_iters
lr_factor = args.lr_decay_target_ratio ** (1 / args.n_iters)
print("lr decay", args.lr_decay_target_ratio, lr_decay_iters)
grad_vars = tensorf.get_optparam_groups(args.lr_init, args.lr_basis)
optimizer = torch.optim.Adam(grad_vars, betas=(0.9, 0.99))
# ==> Training Utils
# ================================
PSNRs, PSNRs_train, PSNRs_test = [],[0],[0]
history = defaultdict(list)
run_tic = time.time()
pbar = tqdm(
range(args.n_iters),
miniters=args.progress_refresh_rate,
file=sys.stdout,
)
# ==> Data preparation
# ================================
allrays, allrgbs = train_dataset.all_rays, train_dataset.all_rgbs
if not ndc_ray: allrays, allrgbs = tensorf.filtering_rays(allrays, allrgbs, bbox_only=True)
trainingSampler = SimpleSampler(allrays.shape[0], args.batch_size)
# Start training
for iteration in pbar:
ray_idx = trainingSampler.nextids()
rays_train, rgb_train = allrays[ray_idx], allrgbs[ray_idx].to(device)
# ==> Get mask
# ================================
ratio = args.mask_ratio_list[0]
if args.free_reg:
free_maskes = get_free_mask(
pos_bl = tensorf.pos_bit_length,
view_bl = tensorf.view_bit_length,
fea_bl = tensorf.fea_bit_length,
den_bl = tensorf.density_n_comp,
app_bl = tensorf.app_n_comp,
using_decomp_mask = args.free_decomp,
total_step = args.n_iters,
step = iteration,
ratio = ratio,
device = device
)
else:
free_maskes = get_free_mask()
# ==> Render from grid
# ================================
(rgb_map, alphas_map, depth_map, weights, uncertainty, num_samples) = renderer(
rays_train,
tensorf,
free_maskes,
chunk=args.batch_size,
N_samples=nSamples,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device,
is_train=True,
)
# ==> Loss
# ================================
total_loss = loss = torch.mean((rgb_map - rgb_train) ** 2)
if Ortho_reg_weight > 0 and 'VM' in args.model_name:
loss_reg = tensorf.vector_comp_diffs()
total_loss += Ortho_reg_weight * loss_reg
summary_writer.add_scalar(
"train/reg", loss_reg.detach().item(), global_step=iteration
)
if L1_reg_weight > 0:
loss_reg_L1 = tensorf.density_L1()
total_loss += L1_reg_weight * loss_reg_L1
summary_writer.add_scalar(
"train/reg_l1",
loss_reg_L1.detach().item(),
global_step=iteration,
)
if TV_weight_density > 0:
TV_weight_density *= lr_factor
loss_tv = tensorf.TV_loss_density(tvreg) * TV_weight_density
total_loss = total_loss + loss_tv
summary_writer.add_scalar(
"train/reg_tv_density",
loss_tv.detach().item(),
global_step=iteration,
)
if TV_weight_app > 0:
TV_weight_app *= lr_factor
loss_tv = tensorf.TV_loss_app(tvreg) * TV_weight_app
total_loss = total_loss + loss_tv
summary_writer.add_scalar(
"train/reg_tv_app",
loss_tv.detach().item(),
global_step=iteration,
)
optimizer.zero_grad()
total_loss.backward()
optimizer.step()
loss = loss.detach().item()
PSNRs.append(-10.0 * np.log(loss) / np.log(10.0))
summary_writer.add_scalar(
"train/PSNR", PSNRs[-1], global_step=iteration
)
summary_writer.add_scalar(
"train/mse", loss, global_step=iteration
)
# ==> Update learning rate
# ================================
for param_group in optimizer.param_groups:
param_group["lr"] = param_group["lr"] * lr_factor
# ==> Print loss
# ================================
if iteration % args.progress_refresh_rate == 0:
step_toc = time.time()
pbar.set_description(
f"Iteration {iteration:05d}:"
+ f" train_psnr = {float(np.mean(PSNRs)):.2f}"
+ f" test_psnr = {float(np.mean(PSNRs_test)):.2f}"
+ f" mse = {loss:.6f}"
+ f" elapsed_time = {step_toc - run_tic:.2f}"
)
PSNRs = []
# ==> PSNRs metric
# ================================
if iteration % args.train_vis_every == 0:
if iteration % args.vis_every == 0:
PSNRs_test = PSNRs_calculate(
tensorf,
test_dataset,
renderer,
chunk=args.batch_size,
N_samples=nSamples,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device
)
history['iteration'].append(iteration)
history['train_psnr'].append(round(float(np.mean(PSNRs_train)), 2))
history['test_psnr'].append(round(float(np.mean(PSNRs_test)), 2))
history['mse'].append(round(loss, 5))
# ==> Save rendered image
# ================================
save_rendered_image_per_train(
train_dataset = train_gift_data,
test_dataset = test_gift_data,
tensorf = tensorf,
renderer = renderer,
step = iteration,
logs = history,
savePath = f'{logfolder}/gif/',
chunk = args.batch_size,
N_samples = nSamples,
white_bg = white_bg,
ndc_ray = ndc_ray,
device = device
)
PSNRs_train = []
return
# ==> Update alphaMask list
# ================================
if iteration in update_AlphaMask_list:
if (reso_cur[0] * reso_cur[1] * reso_cur[2]) < 256**3:
reso_mask = reso_cur
new_aabb = tensorf.updateAlphaMask(tuple(reso_mask), free_maskes['decomp']['den'])
if iteration == update_AlphaMask_list[0]:
tensorf.shrink(new_aabb)
# Filter rays outside the bbox
if not ndc_ray and iteration == update_AlphaMask_list[1]:
allrays, allrgbs = tensorf.filtering_rays(allrays, allrgbs)
trainingSampler = SimpleSampler(
allrgbs.shape[0], batch_size
)
if iteration in upsamp_list:
if len(upsamp_list) == len(mask_ratio_list):
ratio = mask_ratio_list[upsamp_list.index(iteration)]
n_voxels = N_voxel_list.pop(0)
reso_cur = N_to_reso(n_voxels, tensorf.aabb)
nSamples = min(nSamples, cal_n_samples(reso_cur, step_ratio))
tensorf.upsample_volume_grid(reso_cur)
if lr_upsample_reset: lr_scale = 1 # 0.1 ** (iteration / args.n_iters)
else:lr_scale = lr_decay_target_ratio ** (iteration / n_iters)
grad_vars = tensorf.get_optparam_groups(lr_init * lr_scale, lr_basis * lr_scale)
optimizer = torch.optim.Adam(grad_vars, betas=(0.9, 0.99))
if iteration in save_ckpt_every:
tensorf.save(f'{logfolder}/{iteration//1000}k_{expname}.th')
return
tensorf.save(f'{logfolder}/final_{args.expname}.th')
elapsed_time = time.time() - run_tic
np.savetxt(f"{logfolder}/training_time.txt", np.asarray([elapsed_time]))
print(f"Total time {elapsed_time:.2f}s.")
if args.render_train:
os.makedirs(f"{logfolder}/imgs_train_all", exist_ok=True)
train_dataset = dataset(args.datadir, split="train", downsample=args.downsample_train, is_stack=True)
PSNRs_test = evaluation(
train_dataset,
tensorf,
renderer,
f"{logfolder}/imgs_train_all/",
N_vis=-1,
N_samples=-1,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device,
)
print(f"======> {args.expname} test all psnr: {np.mean(PSNRs_test)} <========================")
if render_test:
os.makedirs(f"{logfolder}/imgs_test_all", exist_ok=True)
PSNRs_test = evaluation(
test_dataset,
tensorf,
renderer,
f"{logfolder}/imgs_test_all/",
N_vis=-1,
N_samples=-1,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device,
)
summary_writer.add_scalar("test/psnr_all", np.mean(PSNRs_test), global_step=iteration)
print(f"======> {args.expname} test all psnr: {np.mean(PSNRs_test)} <========================")
if args.render_path:
c2ws = test_dataset.render_path
print("========>", c2ws.shape)
os.makedirs(f"{logfolder}/imgs_path_all", exist_ok=True)
evaluation_path(
test_dataset,
tensorf,
c2ws,
renderer,
f"{logfolder}/imgs_path_all/",
N_vis=-1,
N_samples=-1,
white_bg=white_bg,
ndc_ray=ndc_ray,
device=device,
)
np.savez(f"{logfolder}/history.npz", **history)
create_gif(f"{logfolder}/gif/plot/vis_every", f"{logfolder}/gif/training.gif")
return f'{logfolder}/final_{args.expname}.th'
if __name__ == '__main__':
torch.set_default_dtype(torch.float32)
torch.manual_seed(20211202)
np.random.seed(20211202)
ckpt_path = reconstruction()
"""args = config_parser()
if args.export_mesh:
export_mesh(args, args.ckpt)
if args.render_only and (args.render_test or args.render_path or args.render_train):
render_test(args)
elif args.config:
ckpt_path = reconstruction(args)
export_mesh(args, ckpt_path)
import shutil
shutil.copy"""